Real-time imaging method

Abstract

The invention relates to a real-time imaging method, comprising the following steps: acquiring an original actual series [mu], [mu] = ([mu] 0,..., [mu] n-1) of n measurements according to each given time interval, the n measurements being consecutive in series in a predefined order, each series of n measurements allowing, by processing these measurements, a series of n measurements, each series of n measurements being selected from the group consisting of n measurements; generating at least one raw result X = G (mu) at a given frequency f; generating artificial series from a set of measurements originating from an actual series [mu] with a given rank, and from at least one actual series with a different rank, in particular the next rank, and from permutations P of measurements within each artificial series, so as to observe a predefined order of n measurements within each series; a stream of results is generated at a frequency greater than f from the original results and from results generated from the artificial series between the actual series.

Description

Technical Field

[0001] The present invention relates to the field of polarimetric imaging and more particularly, but not exclusively, to polarimetric imaging as applied to colposcopy. Background Art

[0002] Colposcopy involves examining the cervix using a device called a colposcope. A colposcope allows for remote viewing of the cervix without any contact between the optics or other components of the colposcope and the tissue being examined.

[0003] A colposcope includes a lighting system for illuminating and remotely viewing the cervix through a speculum inserted into the vagina.

[0004] Illumination is conventionally achieved using white light. A green filter can be added to the lighting system and optionally used to highlight the areas most absorbed by hemoglobin. The optical system of the colposcope can be connected to a color camera for recording color images and video on a computer's hard drive (or monochrome images after using a green filter). In addition, images of the cervix can be displayed in real time on a computer screen. In some cases, the optical system of the colposcope is also connected to an eyepiece that allows the user to directly observe the cervix using, for example, stereoscopic vision.

[0005] The colposcope may comprise a head mounted on an articulated arm and may comprise two eyepieces, more specifically a left eyepiece and a right eyepiece, connected to two corresponding light inlet ports present on the front of the head, directed towards the area to be examined, for directly viewing the area of ​​interest, and may comprise, next to the two inlet ports, a light outlet port for illuminating the area of ​​interest; the light being generated, for example, by a lamp arranged outside the head and routed to the head via a light guide comprising a bundle of silicon optical fibers.

[0006] Furthermore, Mueller polarimetric imaging involves measuring the Mueller matrix of a target sample and can provide various useful information about the properties of the sample, particularly by allowing analysis of its optical anisotropy and its light scattering properties.

[0007] Mueller polarimetry imaging is of interest for in vitro and in vivo studies of various biological tissues and in particular the uterine cervix.

[0008] Patent EP 1 738 682 describes components for realizing polarimetric images in colposcopy.

[0009] Indeed, the knowledge provided by polarimetric information can improve the quality of medical diagnostics for various types of conditions and in particular for the early detection of cervical cancer.

[0010] In particular, Mueller polarimetric imaging is a technique that allows for a complete polarimetric characterization of a sample by measuring its Mueller matrix.

[0011] A Mueller polarimeter generally consists of a light source, a polarization state generator (PSG), a polarization state analyzer (PSA), and a detector.

[0012] Mueller polarimeter can provide single-shot measurements. In this case, a photodetector can be used as the detector. In the case of a single-shot Mueller polarimeter, a single Mueller matrix is ​​measured.

[0013] However, a Mueller polarimeter can advantageously be used as an imager. In this case, a CCD or CMOS camera is used as a detector. Each pixel of the camera is equivalent to a photodetector of a single-shot Mueller polarimeter. In the case of an imaging Mueller polarimeter, the Mueller matrix is ​​measured for each pixel. Mueller polarimetric imaging generally requires the acquisition of multiple intensity images for measuring the Mueller matrix of the sample. This is a technique that can be slow, especially if the intensity images required to obtain the Mueller matrix are acquired continuously. In the case of an imaging Mueller polarimeter, the luminescence intensity is measured simultaneously for all pixels of the camera used. For each pixel, the luminescence signal is converted into photoelectrons. Among other factors, the modality with which the electrical signal generated by the photoelectrons is processed and the number of pixels in the selected region of interest (ROI) of the sensor determine the number of images acquired per second ("frames per second" or FPS).

[0014] If multiple wavelengths have to be acquired to explore biological tissue at different depths, Mueller polarimetry imaging requires even more time if these wavelengths are acquired sequentially, which is the case for most currently used multispectral Mueller polarimeters.

[0015] However, multispectral analysis is crucial for exploring biological tissue at varying depths. Indeed, shorter wavelengths in the visible spectrum, such as those corresponding to blue and green, are significantly absorbed by hemoglobin and primarily allow exploration of the superficial layers of biological tissue. In contrast, red / near-infrared light, which is absorbed to a much lesser extent by hemoglobin, allows exploration of biological tissue at greater depths. For wavelengths in the visible / near-infrared spectrum, the penetration length of light in biological tissue depends on the microscopic absorption and scattering properties of the tissue in question.

[0016] Generally speaking, the light penetration length in the red / near infrared portion of the electromagnetic spectrum increases with wavelength.

[0017] There are many different types of Mueller polarimeter in the literature. The most widely used is the time-series Mueller polarimeter, which sequentially acquires the intensity images required to obtain the Mueller matrix.

[0018] Mueller polarimetry imaging is performed in three main steps:

[0019] 1) Measuring the intensity matrix (according to the established notation, B=AMW). This matrix can be acquired multiple times in order to increase the signal-to-noise ratio by an averaging process, thereby improving the reliability of the measurement. This can be called a measurement step;

[0020] 2) Obtain the Mueller matrix M = A -1 BW -1 , which can be called a processing step;

[0021] 3) determining the relevant polarimetric parameters from M, which may be referred to as a post-processing step,

[0022] For example, different algebraic methods are used, such as Mueller matrix decomposition.

[0023] For biomedical applications, step 1) must be completed within a second or so at most to reduce the effects of blurring caused by involuntary patient movement (breathing, heartbeat, etc.) during measurement. Steps 2) and 3) should ideally be completed very quickly to render a useful image to the practitioner as soon as possible after the measurement, with a maximum delay of a few seconds.

[0024] Step 2) can generally be fast and fairly easy to implement, especially if the matrix A -1 and W -1 Already pre-calculated words.

[0025] In contrast, step 3) can be rather slow and require high computing power, especially if it requires, for example, computing the eigenvalues ​​of the matrix M, which may often be the case with Mueller matrix decomposition or other algebraic processing operations.

[0026] To determine the Müller matrix of a sample using a time-series Müller polarimeter, a series of luminous intensity measurements is performed using PSG and PSA. For a 4×4 Müller matrix, at least 16 measurements are required, allowing the polarization of the light sent to the sample to be modulated and the polarization of the light rendered by the sample to be analyzed. Various combinations of PSG and PSA allow the acquisition of the 16 intensity images required to obtain the Müller matrix of the sample.

[0027] For 4×4 Mueller polarimetry imaging (full Mueller polarimetry imaging), at least 16 intensity images need to be acquired, which are grouped into an actual 16-component intensity matrix B.

[0028] The use of a polarization camera allows the acquisition of 12 intensity images, which grants access to a 3×4 Mueller matrix. For example, for biological tissues with specific polarimetric properties, it is possible to start from the measured 3×4 Mueller matrix and use algebraic calculations to reduce it to a 4×4 Mueller matrix.

[0029] However, several factors limit the practical use of polarization cameras. Polarization cameras are typically monochrome. Therefore, several cameras are required to reconstruct a color image, which significantly increases the system size. Furthermore, images acquired with various cameras are difficult to superimpose pixel by pixel. Furthermore, obtaining pixel-by-pixel superimposable images acquired with various cameras requires very precise optical adjustments, which are quite complex to implement. Furthermore, with these cameras, each pixel is divided into four sub-pixels, two of which correspond to two different linear polarization states. The remaining two pixels generally correspond to the same linear polarization state between them, which is different from the polarization state corresponding to the two first pixels mentioned above. The division of a pixel into four sub-pixels results in a loss of image resolution. This loss of resolution can be restored using digital methods, although this generates artifacts in the image, or it can be restored using relatively complex computational methods, which significantly increases image acquisition time. Finally, because the polarization states of the individual sub-pixels are not completely separated, crosstalk occurs between them.

[0030] For biomedical applications, the matrix B has to be acquired within a maximum of a second or so and the relevant polarimetric parameters calculated during the post-processing step have to be rendered with a delay of a maximum of several seconds.

[0031] In order to reduce the influence of blurring caused by the patient's involuntary movement during measurement, it is necessary to acquire the intensity matrix B within about one second. Within one second, the matrix B can be acquired multiple times using an averaging process to increase the signal-to-noise ratio.

[0032] Therefore, rendering the most relevant polarimetric parameters from the Mueller matrix within seconds after post-processing using, for example, a decomposition of the measured Mueller matrix is ​​crucial to allow the practitioner to obtain useful information very quickly during a consultation.

[0033] An ideal scenario would involve being able to render the measured Mueller matrix in real time, or even the related polarimetric parameters in real time (after post-processing the Mueller matrix).

[0034] However, during clinical examinations, it is difficult for practitioners to obtain a stream of images rendered to the user in substantially real time at a frequency sufficient for comfortable viewing.

[0035] The first limitation is the acquisition speed of the intensity images required to obtain the intensity matrix B and thus the Mueller matrix M.

[0036] The second limitation is related to the use of post-processing procedures to obtain the main polarimetric parameters of M. For example, most Mueller matrix decompositions require computing the eigenvalues ​​of the Mueller matrix M, which represents a computational process that can be difficult to parallelize and perform in real time.

[0037] In practice, in conventional white-light imaging systems, for smooth real-time tracking of images, if the target moves very slowly, the stream rate should generally be at least 8 FPS. Therefore, in the case of Mueller polarimetry imaging, if 16 measurements are to be taken between two displayed images, the number of images to be acquired and processed is at least 8 * 16 = 128, which can be a very high frequency for high-resolution images.

[0038] Some polarimetric imaging methods allow for a reduction in the number of measurements, but as a result do not allow for the acquisition of a full Mueller polarimetric image (4×4). Other methods are based on relatively complex specialized sensors, such as polarization cameras, which allow for faster acquisition, but in their current state may have other significant limitations as described above.

[0039] If a multispectral approach is required and the various wavelengths are acquired sequentially, real-time imaging becomes difficult to achieve. One possibility would be to use several monochrome CCD or CMOS cameras simultaneously, coupled to optical filters for selecting the wavelength range of interest. This would make the polarimetry system bulky and difficult to integrate compactly into existing imaging systems or into completely new systems for medical use.

[0040] Therefore, in order to adapt Mueller polarimetry for example for colposcopy or to apply it to other optical imaging techniques for biomedical applications such as endoscopy or even microscopes (or exoscopes) for neurosurgery, several problems arise.

[0041] The first challenge is to achieve real-time Mueller polarimetry imaging. This involves rendering the Mueller matrix in real time after the measurement and processing steps, and even more preferably, the most relevant polarimetry parameters after the step of post-processing the Mueller matrix, so as to allow practitioners to immediately (e.g., during or within seconds of a patient visit) access information related to the microstructure of the tissue.

[0042] The second challenge involves performing step 1) above simultaneously for several wavelengths in the visible / near-infrared spectrum to allow practitioners to explore the microstructure of tissue at different depths.

[0043] A third challenge also involves providing real-time reference color imaging, which is necessary to clearly identify the analyzed area with polarimetry, thereby allowing the practitioner to have well-known spatial markers, which is not possible with monochrome luminous intensity images or by directly displaying polarimetry images.

[0044] Another challenge relates to the possibility of pixel-by-pixel superposition of various types of images (such as, for example, color images, luminous intensity images of various selected wavelength ranges, and polarimetric images of various selected wavelength ranges) in all possible combinations, which is crucial for efficient image analysis using, for example, image processing algorithms or learning algorithms. This step can, among other things, allow: 1) selection of the most relevant polarimetric and non-polarimetric parameters; 2) determination of the most relevant combination of polarimetric and non-polarimetric parameters for diagnosis; 3) combining polarimetric images of interest (e.g., elements of a Mueller matrix and / or polarimetric parameters obtained after algebraic manipulation operations such as Mueller matrix decomposition) and non-polarimetric images (e.g., color images and / or monochromatic intensity images) at various wavelengths in a single image or a limited number of images using, for example, various types of image processing algorithms in order to provide the practitioner with simplified and optimized information for diagnosis that cannot be directly observed in conventional images or in the initial uncombined polarimetric images; and 4) combining the polarimetric images obtained for different wavelength ranges in order to obtain a colored polarimetric image.

[0045] Another challenge may involve rendering the combined image substantially in real time. Finally, a last challenge relates to having a Mueller polarimeter that is very compact and ergonomically easy to adapt to various existing imaging systems (such as, for example, colposcopes, endoscopes, or microscopes (or exoscopes) for neurosurgery) or easy to use to create new ergonomic and compact imaging systems for medical practice depending on the intended application. Summary of the Invention

[0046] Therefore, there is a need to benefit from a high performance polarimetric imaging system, in particular a polarimetric colposcope, in order to allow multispectral and / or color visualization, as well as comfortable multispectral polarimetric and / or color visualization in the examined area, in particular in real time.

[0047] More generally, there is a need to provide a solution for increasing the frequency of images for any imaging technique (such as Mueller polarimetry) that requires a large number of measurements before an image can be generated, so that the practitioner's visualization of the image during the examination is more streamlined, thereby facilitating decision-making, and / or reducing the duration of the examination, etc.

[0048] In addition to the above advantages, there are also advantages:

[0049] - Facilitates post-processing of the Mueller matrix in order to reduce the demands on computing resources and the time required to calculate polarimetric images;

[0050] - facilitating the production of imaging systems, in particular colposcopes, which are capable of performing both multispectral and / or color imaging and multispectral and / or color polarimetric imaging, and in particular allowing conventional viewing systems, such as conventional colposcopes, to be easily converted into viewing systems capable of multispectral and / or color imaging and multispectral and / or color polarimetric imaging;

[0051] - Improvements in the lighting system of colposcopes, in particular to improve the quality of the images;

[0052] - generally improving the ergonomics of viewing systems such as colposcopy systems; and / or

[0053] - Facilitates the setup of the system, in particular in order to perform calibration.

[0054] The present invention aims to address all or some of the requirements set out above.

[0055] Multispectral polarimetry and / or color imaging

[0056] According to a first aspect of the present invention, the object of the present invention is a multispectral polarimetry and / or color imaging system, in particular a polarimetric colposcopic imaging system, comprising:

[0057] - an illumination system comprising at least one light emitting source, this illumination system emitting in at least two, preferably at least three, spectral bands;

[0058] - a polarization state generator (PSG), deployed downstream of the light source and upstream of the target to be imaged;

[0059] - a polarization state analyzer (PSA), deployed downstream of the target to be imaged;

[0060] - a multi-sensor camera comprising at least two, preferably at least three, sensors for recording at least two, preferably at least three, images in said spectral bands, respectively.

[0061] Polarimetric imaging systems can be wide-field (macroscopic) or microscopic and can operate in free space or not, by transmission or reflection. It is also possible to have an endoscope between the target and the PSA and a liquid guide between the PSG and the target, as described in EP 2021 / 052647. Preferably, it is wide-field and operates by reflection, especially in free space (such as in colposcopy). The imaging system can also be used for other biomedical applications, for example, brain surgery or endoscopy, as well as non-biomedical applications, for example, in the fields of cosmetics or microelectronics.

[0062] The use of a luminous source emitting in several spectral bands, combined with the use of a multi-sensor camera, allows for the generation of images in various wavelength ranges that are perfectly superimposable, since the individual sensors of the camera, by construction, produce perfectly superimposable images, i.e., for all pixels of the image, the same point on the image corresponds to the same pixel on each sensor. In particular, since the same camera is used to produce both the color image and the polarimetric image, and the system produces images that are the same size and positioned identically on each sensor, it is particularly possible to limit parallax effects between the various types of images produced by the camera, for example, between a color image (a conventional RGB image) and a polarimetric image, thereby allowing their pixel-by-pixel superposition. Parallax effects can also be limited by combining several cameras, but aligning several cameras can be complex and time-consuming. The previously proposed solution allows for more easily limiting parallax effects by using a single camera.

[0063] Preferably, there are three spectral bands and for example 445 nm to 475 nm for the first band, 510 nm to 550 nm for the second band and 600 nm to 660 nm for the third band, for example centered at 460 nm, 530 nm and 630 nm respectively.

[0064] Preferably, the camera is a dual-CCD, dual-CMOS, triple-CCD, triple-CMOS, quad-CMOS, or 4-CCD camera, preferably a triple-CCD, triple-CMOS, quad-CMOS, or 4-CMOS camera. Also preferably, three spectral bands are present, located in the red or near-infrared, green, and blue bands, respectively. For example, for a 2-CCD camera, the spectral bands are preferably visible light and near-infrared; for a 3-CCD camera, the spectral bands are preferably blue, green, and red / near-infrared; and for a 4-CCD camera, the spectral bands are preferably blue, green, red, and near-infrared.

[0065] This allows, for example, the use of fast, high-resolution cameras manufactured on a large scale. Multi-sensor cameras provide grayscale, independent intensity images for each sensor, which can be used to measure the coefficients of the intensity matrix B in the case of Müller polarimetry and combined to reconstruct a color image of the observed area.

[0066] The camera can include at least two, and preferably three, dichroic prisms for separating the wavelengths toward the individual sensors. For example, a triple CCD camera includes three dichroic prisms for separating the wavelengths toward the individual sensors. In a specific manner known per se, the blue detection sensor can be attached to a first prism, the red detection sensor can be attached to a second prism, which can itself be attached to the first prism, and the green detection sensor can be attached to a third prism, which can itself be attached to the second prism. Other arrangements are possible, and cameras of this type are well known.

[0067] Preferably, the polarization state generator (PSG) comprises an electrically controllable liquid crystal polarization modulator, which offers relatively fast switching, good compactness, suitable bandwidth, and does not disrupt the ability of images to be superimposed. Liquid crystals also allow for full-field polarimetric imaging for analyzing macroscopic-scale surfaces, as well as microscopic polarimetric imaging.

[0068] As a variant, the polarization state generator comprises, for example, a rotating wave plate and a rotating polarizer and / or a coupled system of a rotating wave plate and a rotating polarizer. As a variant, the polarization state generator may comprise a system coupled with a rotating wave plate and a fixed polarizer, or even a system coupled with a fixed wave plate and a rotating polarizer, or even a system coupled with a rotating retardation plate and a rotating polarizer.

[0069] Polarization state generators can also be generated using spatially separated polarization systems (polarizers, retardation plates, etc.) that allow the generation of various polarization states of light.

[0070] Polarization state generators can also be generated using spatially separated polarization systems (polarizers, hysteresis plates, etc.), which allow the generation of various polarization states of light. Spatially separated polarization systems can also be fixed or can be temporally modulated using a liquid crystal system or a rotating system. In this configuration, the emission beam can be deflected using a fast switching system to successively pass through the various polarization systems before illuminating the target. Similarly, the polarization state analyzer preferably includes an electrically controllable liquid crystal depolarization modulator. As a variant, the polarization state analyzer includes, for example, a rotating wave plate according to all the combinations described above for the polarization state generator, as well as a rotating polarizer or even a system of coupled wave plates and polarizers. Preferably, the liquid crystal polarization modulator is a ferroelectric liquid crystal modulator, which allows high switching frequencies with a control voltage of several volts. As a variant, the liquid crystal polarization modulator is a nematic liquid crystal modulator. Polarization state analyzers can also be generated using spatially separated polarization systems (polarizers, hysteresis plates, etc.), which allow the generation of various polarization states of light. Like the polarization state generator, the polarization system is spatially separated and can also be fixed or can be temporally modulated using a liquid crystal system or a rotating system. In this configuration, the emission beam can be deflected using a fast switching system so that it passes through various polarization systems successively before illuminating the target (and in particular before entering the detector).

[0071] The light source is preferably a white light source, in particular a xenon lamp. Such a lamp allows emission across a broad spectrum and allows both good multispectral and / or color images to be produced, as well as multispectral polarimetric and / or color images. As a variant, the light source can be a halogen lamp, at least one LED, or any other type of light source, either incoherent or coherent (laser).

[0072] When using a xenon lamp, the lighting system can include a single-band, dual-band, preferably a triple-band dichroic filter, which is located downstream of the light source, placed before the target to be analyzed, or placed after the target to be analyzed and before the detector. This filter can block, for example, UV and IR, and allow white light (especially the red, green, blue bands in the visible spectrum) to pass. This filter can be removable, in particular mounted on a filter wheel, so as to allow, for example, acquisition in IR, near IR or even near UV. As a variant, depending on the type of light source used, the lighting system can include a dual-band, preferably a triple-band filter, or at least one single-band filter. For example, a quad-band filter can be used with a 4CMOS or 4CCD camera, for example to allow simultaneous acquisition of images in the blue, green, red and near-infrared bands.

[0073] It is also possible not to use spectral filters.For example, the lighting system may comprise several light emitting sources each emitting in a spectral band of interest, such as blue, green and red, or blue and yellow LEDs.

[0074] The imaging system may comprise a polarimetry system according to another aspect of the invention, as defined below.

[0075] The imaging system may comprise a filter wheel arranged directly upstream of the camera, optionally together with a linear polarization filter of an analyzer, as described below. As a variant, the filter wheel may also be arranged directly downstream of the light source.

[0076] The filter wheel can support a three-band filter as mentioned above, for example to allow the camera to acquire images with wavelengths in a spectral range centered around 460 nm, 530 nm, and 630 nm; and one or more monochromatic dichroic filters, for example to allow the acquisition of images in a spectral range around 650 nm and 700 nm. A sensor designed for imaging the red band can also allow imaging in the near infrared band.

[0077] The filter wheel can select different spectral ranges in the blue, green, and red / near infrared bands by changing the triple-band filters, removing filters, or replacing the type of filters used (e.g., replacing the triple-band filters with at least one monochromatic filter). The filter wheel can allow some wavelength ranges to be explored in more detail.

[0078] The imaging system may comprise a processor for generating a Mueller polarimetric image of the target in each spectral band. A Mueller polarimetric image is understood to mean an image comprising at least one item of information extracted from the Mueller matrix, for example, a polarimetric property (such as polarization birefringence, polarization, birefringence, depolarization, etc.), as well as any information derived at least in part from one of these properties. In the case of colposcopy for analyzing the cervix, knowledge of the birefringence and depolarization properties is particularly useful for diagnosis. Polarimetric parameters particularly useful for diagnosis when used in colposcopy may be, inter alia, the linear phase lag, the uncertainty level of the azimuth of the slow axis (or fast axis) of the linear phase lag, the combined image of the linear phase lag and the azimuth of the slow axis (or fast axis) of the linear phase lag, depolarization, and the linear correlation between depolarization and the linear phase lag, in particular at 530 nm.

[0079] The processor may be configured to at least partially superimpose an image containing at least one item of polarimetric information on an image corresponding to a non-polarimetric observation of the camera (eg, a primary color image). This polarimetric information may be displayed in false color.

[0080] The system can be designed to carry out, on a given sample, a series of successive actual measurements, in particular with each sensor of the camera and simultaneously for all the various sensors, these measurements corresponding to obtaining at least 9 (3×3 Müller polarimetry), in particular 12 (3×4 Müller polarimetry), preferably 16 (4×4 Müller polarimetry) coefficients of the intensity matrix B,

[0081] The processor is designed to:

[0082] - generating a series of n additional measurements (also called "artificial" series) from a series μ originating from measurements of a given rank and from a set of measurements originating from at least one series of a different rank (in particular the next rank) and from a permutation P(μ) of the measurements within each artificial series, so as to observe a predefined order of the n measurements within each series; then

[0083] - generating, from the polarimetric images produced by the real measurement series and the polarimetric images produced by the artificial series between the real series, a sequence of polarimetric images having a frequency greater than that of the real measurement series that would have been allowed in the absence of the artificial measurement series.

[0084] The processor may be designed to perform decomposition of the Mueller matrix into a non-depolarized component and a depolarized component. The depolarized component and the non-depolarized component may take a variety of forms.

[0085] Therefore, the processor can be designed to perform an additive solution on the Mueller matrix M, of the form M = qM nd +pM d , where M nd is the non-depolarized component and M d is the depolarization component. The parameters q and p are the components of the Mueller matrix M nd and M d The weight of .

[0086] According to this model, the measured intensity matrix B can be written as the sum of the polarization contribution and the non-depolarization contribution.

[0087] B=B nd +B d =A(qM nd +pM d )W

[0088] Then

[0089] M=qM nd +pM d =A -1 (B nd +B d )W -1

[0090] For a 4×4 Mueller matrix, then

[0091] B=AMW,

[0092] in

[0093]

[0094] as well as

[0095] The measurement is generated using four Stokes vectors The four independent states of the polarization state generator PSG are described and generated by four Stokes vectors Describe the analyzed states obtained by the polarization state analyzer PSA.

[0096] This increases the image frequency, improving viewing comfort and enabling real-time polarimetric imaging, i.e., a stream of at least eight polarimetric images per second. Consequently, as practitioners change the observed area, they can quickly benefit from the corresponding polarimetric images. Increasing the image frequency allows the number of images per second to be artificially increased without changing the exposure time required for acquisition, thus offering the possibility of acquiring images in real time. If the observed object is stationary or moving slowly, a frequency of eight images per second is sufficient. If the object is moving more rapidly, a higher image frequency may be required.

[0097] The processor may be designed to spatially recalibrate the images corresponding to the measurements before performing the permutation P, so as to take into account possible displacements of the observed area within the field of view of the camera.

[0098] In this way, the risk of blurring the polarimetric image is limited and the accuracy of the polarimetric information is increased.

[0099] The processor can be designed to perform a decomposition of the Mueller matrix into a non-depolarized component and a depolarized component. The depolarized component and the non-depolarized component can take several forms. Thus, the processor can be designed to perform an additive decomposition of the Mueller matrix M, which is of the form M = qM nd +pM d , where M nd is the non-depolarized component and M d is the depolarization component. The parameters q and p are the weights of the two components of the Mueller matrix.

[0100] In particular, M nd The Mueller matrix of the linear phase lag can be given by:

[0101]

[0102] And Md The Mueller matrix of a pure depolarizer can be given by:

[0103]

[0104] θ is the azimuthal orientation of the fast axis (or slow axis) in degrees and δ is the linear phase lag in degrees (ranging from 0° to 180°);

[0105] in

[0106]

[0107]

[0108] The Mueller matrix M can be expressed relative to its unnormalized and unpolarized intensity coefficients Normalized and can be written as:

[0109]

[0110] in

[0111]

[0112] So the Mnormalised matrix can be written as:

[0113]

[0114] The coefficient m ij (i, j = 1, 2, 3, 4) are the coefficients of the normalized Mueller matrix, is the unnormalized and unpolarized intensity coefficient of M, and the other coefficients m ij (i, j = 1, 2, 3, 4) relative to is normalized.

[0115] The processor may be designed to calculate the depolarization by performing the following calculation:

[0116]

[0117] The system may be designed to generate a parallel display of at least one non-polarimetric image, in particular a color image of the observed area, and at least one polarimetric image in at least one spectral band.

[0118] Another object of the present invention is a method for training an artificial intelligence system, for example comprising at least one convolutional neural network, wherein the artificial intelligence system can receive as input non-polarimetric color and / or multispectral images and polarimetric color and / or multispectral images.

[0119] The fact that these images originate from a multi-sensor camera avoids additional spatial recalibration operations for images acquired simultaneously by various sensors with different wavelengths and reduces computation time, which leaves more resources available for training. The images acquired by the various sensors can be superimposed pixel by pixel.

[0120] Increase image frequency

[0121] According to another aspect of the invention, preferably in combination with the above, but generally applicable to other types of imaging, the invention has as its object a method for real-time imaging comprising the following steps:

[0122] - obtaining at each given time interval an original actual series μ of n measurements following one another in a series in a predefined order, μ=(μ0, ..., μ n-1 ), wherein each series of n measurements allows, by processing these measurements, to generate at least one raw result X=G(μ) at a given frequency f, where G is the function generating the result from the measurements;

[0123] - generating an artificial series from a set of n measurements, the set of measurements originating from a real series μ with a given rank and from at least one real series with a different rank (in particular the next rank), and from a permutation P of the measurements within each artificial series so as to observe a predefined order of the n measurements within each series;

[0124] - Generate a stream of results whose frequency is greater than f from the original results and from the results generated by artificial series inserted between the original real series.

[0125] This imaging method is advantageously applied in polarimetric imaging, such as polarimetric colposcopy, and preferably in multispectral and / or color polarimetric colposcopy as defined above.

[0126] This method can be applied to colposcopy, microscopy, exoencephaloscopy, endoscopy, preferably to polarimetric colposcopy, and more preferably to multispectral polarimetric colposcopy.

[0127] The result is then the Mueller matrix or a physical property calculated from this matrix. This series of measurements is a polarimetric measurement.

[0128] According to this aspect, the invention allows to artificially increase the flow of images and allows or improves real-time imaging regardless of the number of measurements to be performed and the calculations required to generate the desired image.

[0129] In other words, according to this aspect, the invention allows to artificially increase the number of images per second while maintaining a fixed number of acquisitions.

[0130] "Real time" is understood to mean a relatively short delay between acquisition and visualization, for example, 2 or 3 seconds, and compatible with visualization of information generated by the practitioner during the examination; for example, the delay between the time a measurement is taken and the time the information resulting from those measurements is displayed is less than or equal to 2 seconds, preferably 1 second. The resulting stream of results may be greater than 8 per second.

[0131] The imaging to which this method is applied is advantageously polarimetric imaging and the series of n measurements can correspond to the measurement of the coefficients of the intensity matrix B for calculating the Mueller matrix.

[0132] In particular, the series of measurements may correspond to the acquisition of at least 9, in particular 12, preferably 16 coefficients of the intensity matrix B, for example.

[0133] As indicated above, for a 4×4 Mueller matrix,

[0134] B=AMW,

[0135] in

[0136]

[0137] as well as

[0138] The measurement is generated using four Stokes vectors The four independent states of the polarization state generator PSG are described and generated by four Stokes vectors Describe the analyzed states obtained by the polarization state analyzer PSA.

[0139] According to this model, the measured intensity matrix B can be written as the sum of the polarization contribution and the non-depolarization contribution.

[0140] B=B nd +B d =A(qM nd +pM d )W

[0141] M=qM nd +pM d =A -1 (B nd +B d )W -1

[0142] Preferably, as mentioned above, before performing the permutation, the images corresponding to the various intensity measurements representing the components of the matrix B (obtained for the various configurations of PSG and PSA) are spatially recalibrated in order to take into account possible displacements of the area observed in the field of view of the camera. The method for increasing the frequency of the images can be generally applied to systems using, for example, multi-sensor CCD and CMOS cameras, as well as to systems using monochrome or polarization CCD and CMOS cameras.

[0143] Post-processing of the Mueller matrix

[0144] In order to speed up the calculation of the Mueller matrix, specific post-processing can be implemented that, for example, avoids performing a decomposition, known as the Lu-Chipman decomposition, but still allows obtaining close results.

[0145] Therefore, according to another aspect of the present invention, a further object of the present invention is to provide a polarimetric imaging method, independently or in combination with the above, wherein an intensity image is acquired using a polarimetric imaging system (preferably a system as defined above), the polarimetric imaging system comprising an illumination system, a polarization state generator (PSG) placed on the optical path between the illumination system and the area to be observed, a polarization state analyzer (PSA) placed on the optical path between the area to be observed and at least one image acquisition system, and then performing an additive decomposition of the Mueller matrix M, which has the form M=qM nd +pM d , where M nd is the non-depolarized component and M d is the depolarized component. In particular, M nd The Mueller matrix of the linear phase lag can be given by:

[0146]

[0147] And M d The Mueller matrix of a pure depolarizer can be given by:

[0148]

[0149] θ is the azimuthal orientation of the fast axis (in degrees) and δ is the phase lag (in degrees) (ranging from 0° to 180°);

[0150] in

[0151]

[0152]

[0153] The parameters q and p are the weights of the two components of the Mueller matrix.

[0154] The Mueller matrix M can be expressed relative to its unnormalized and unpolarized intensity coefficients Normalized and can be written as:

[0155]

[0156] in

[0157]

[0158] So the matrix Mnormalised can be written as:

[0159]

[0160] Coefficient m ij (i, j = 1, 2, 3, 4) are the coefficients of the normalized Mueller matrix, is the unnormalized and unpolarized intensity coefficient of M, and the other coefficients m ij (i, j = 1, 2, 3, 4) relative to is normalized.

[0161] The processor may be designed to calculate depolarization by performing the following calculations, for example:

[0162]

[0163] This additive decomposition of the Mueller matrix allows the calculations to be parallelized and allows for faster extraction of useful parameters.The Mueller matrix and the associated polarimetric parameters can be obtained for a single wavelength range or for several wavelength ranges simultaneously (especially at least two wavelength ranges, in particular in the visible and infrared bands).

[0164] This method thus allows the simultaneous acquisition of unnormalized and unpolarized intensity coefficients from three Mueller matrices in the spectral range corresponding to the blue, green and red parts of the visible spectrum. to reconstruct images, especially color images, in real time.

[0165] This approach avoids the calculation of the eigenvalues ​​of the Mueller matrix, which is costly in terms of computation time, and allows the extraction of the desired parameters with a much shorter computation time.

[0166] This approach is based on the assumption that in some tissues, such as the cervix, non-depolarizing effects are primarily associated with the most superficial layers of the tissue, whereas depolarizing effects are associated with the bulk of the tissue.

[0167] Observation system with two inlet ports equipped with polarimetry system

[0168] According to another aspect of the invention, independently or in combination with the above content, another object of the present invention is an observation system with two inlet ports, in particular a binocular system, for example, for colposcopy, comprising a head, which includes an optical system having a light outlet port for illuminating the area to be inspected and left and right inlet ports pointing to the area to be observed, the observation system also comprising a polarization measurement system, which includes a polarization state generator arranged in front of the outlet port and a polarization state analyzer, at least some optical elements of the polarization state analyzer are arranged in front of one of the inlet ports, this analyzer including at least one optical element supported by at least one support, which support is only applied to a part of the contour of the optical element.

[0169] This can allow the optical element to have a free edge that partially overlaps the other inlet port. The optical element can only partially overlap the other inlet port without blocking the inlet port with a support that would interfere with observation. For example, the optical element is a liquid crystal polarization modulator.

[0170] This aspect of the invention exploits the fact that the free edge of the optical element, which preferably has a circular profile, although partially overlapping the entrance port, remains substantially transparent in the eyepiece associated with this entrance port due to the focusing distance, which is typically around 20 to 40 cm and is therefore much greater than the distance between the entrance port and the optical element (a few centimeters at most). Furthermore, since the optical elements of the analyzer are relatively transparent, with the possible exception of the linear polarizer, parts of which can be placed in front of the camera, downstream of the deflection prism, towards the eyepiece, these elements do not cause a substantial loss of brightness for the observer.

[0171] It is thus possible to use a conventional colposcope head or the head of another observation system (especially a binocular observation system) with two inlet ports by placing all or part of the polarimetry system in front of it, which limits the manufacturing costs of the system and allows the polarimetry imaging functionality to be easily added to the observation system.

[0172] Each support of an optical element of the analyser may extend in contact with the optical element, for example over an angular range between 180° and 300°, preferably presenting a substantially C-shape opening substantially at 45° in a downwardly inclined direction.

[0173] The polarimetry system may further comprise at least one other support for holding at least one optical element of the polarization state generator placed in front of the light exit port of the colposcope head.

[0174] This further support may present a generally upwardly open C-shape, which limits the volume of the state generator in the vertical direction and avoids impacting the inlet port present near the outlet port on the front surface of the head of the colposcope.

[0175] Each support may comprise a series of parts holding together the optical elements of the polarization state generator (or analyzer), with at least two supports being used to hold two consecutive optical elements sharing an intermediate support part.

[0176] Thus, the polarimetry system can include at least two continuous support parts assembled against one another, each of these parts having a housing for receiving a corresponding optical element on one of its faces, and one of the support parts being used to retain the optical element of the other part. In this way, a relatively compact polarimetry system can be provided that does not further hinder the operator's manipulation of the viewing system head.

[0177] The polarization state generator and the analyzer preferably each comprise similar optical elements placed in an opposite order with respect to the direction of light propagation.

[0178] Therefore, the generator comprises, in the light propagation direction, for example, a linear polarization filter, a quarter-wave liquid crystal polarization modulator QFLC, a half-wave plate QWP and a half-wave liquid crystal polarization modulator HFLC.

[0179] The analyzer may then include a half-wave liquid crystal polarization modulator HFLC, a half-wave plate QWP, and a quarter-wave liquid crystal polarization modulator QFLC in the light propagation direction.

[0180] A polarizer associated with the analyzer is preferably deployed upstream of the camera for collecting the intensity image used to generate the Mueller matrix, after the light returns from the entrance port toward the associated eyepiece, thereby limiting the brightness difference between the images delivered to the observer by the left and right eyepieces, as mentioned above.

[0181] The housing receiving the electrically controllable optical element may comprise at least one channel for the electrical cable. This channel may comprise a channel conforming to the shape of the contour of the element, in particular a semicircular channel.

[0182] The support for the C-shaped optic allows the optic to be manually rotated about its axis before tightening in order to perform orientation adjustments.

[0183] The optical elements of the generator are preferably oriented perpendicularly to the direction of propagation of the light emerging from the head of the viewing system.

[0184] When this light is emitted through a prism at an angle to the viewing direction of the inlet port, the optical elements of the generator are preferably mounted at an angle to the optical elements of the analyzer so that they are oriented perpendicularly to the propagation axis of the light emitted from the outlet port. This allows stray reflections to be limited.

[0185] Preferably, the polarimetry system comprises a fan for blowing air towards the optical elements of the polarization state generator close to the light exit port of the head of the observation system, thereby ensuring better and faster temperature stabilization and limiting corresponding drift.

[0186] The polarization system can comprise a movable filter holder in front of the analyzer for calibrating the system. This filter holder is, for example, fixed to one of the support parts for holding the optical element. The filter holder can be movable in a horizontal direction (generally perpendicular to the viewing direction). The filter holder can adopt, for example, four positions, three of which are used to place a predefined optical element in front of the analyzer and the last position is used to release the field. The displacement of the filter holder is preferably controlled electrically and by the controller mentioned above.

[0187] The system may include a housing that protects the support and the filter holder.

[0188] Another object of the invention is a polarimetry system intended to be attached to the head of an observation system, in particular the head of a colposcope, the head comprising an optical system having a light exit port for illuminating the area to be examined and left and right entry ports directed towards the area to be observed, the polarimetry system further comprising a polarization state generator placed in front of the exit ports and a polarization state analyzer, at least some of the optical elements of which are placed in front of one of the entry ports, this analyzer comprising at least one optical element held by at least one support, the support being applied only to a portion of the contour of the optical element.

[0189] This may allow the optical element to be provided with a free edge that is arranged to partially overlap another inlet port.

[0190] The polarimetry system may include components for attachment to a head of the viewing system.

[0191] These attachment means may comprise screws which are screwed into the frame of the head, for example below the outlet window. As a variant, the attachment is performed in other ways, for example by clamping.

[0192] The polarization system may have all or some of the above mentioned features.

[0193] Compact and modular polarimetry system

[0194] Another object of the present invention is a polarimetry system intended for use in the head of an observation system, in particular a colposcope head, the head comprising an optical system having a light outlet port for illuminating the area to be examined and at least one inlet port, the polarimetry system comprising a polarization state generator (PSG) placed in front of the outlet port and a polarization state analyzer (PSA), at least some of the optical elements of the analyzer being placed in front of the inlet port, wherein at least one of the analyzer and the generator comprises optical elements held by at least one support, each support comprising a series of parts holding together the optical elements of the polarization state generator or analyzer, wherein at least two supports are used to hold two consecutive optical elements that share an intermediate support part, each of these parts preferably having a housing on one surface for receiving the corresponding optical element, and one of the parts is used to retain the optical elements of the other part.

[0195] Such a polarimetry system is particularly compact since it comprises a support part serving as a housing for receiving the optical assembly and a closing cover for the next support part.

[0196] This system is also modular in that it is easy to replace support parts with other support parts.

[0197] It may have any of the features of the viewing system described elsewhere, in particular the colposcope. The polarimetric system may in particular comprise an analyser comprising at least one optical element supported by at least one support member applied only to a portion of the contour of the optical element, thereby providing this optical element with a free edge arranged to partially overlap the other inlet port. The polarimetric system comprises means for attachment to a head of the viewing system described elsewhere.

[0198] Colposcopy lighting system

[0199] According to another aspect of the present invention, independently or in combination with the above, another object of the present invention is a colposcopy system comprising:

[0200] -Lighting system, including:

[0201] ○At least one light source;

[0202] A liquid light guide connected at one end to a light source, preferably having a core diameter of 5 mm or less;

[0203] a colposcope head comprising an optical system having a light inlet port connected to the other end of the liquid light guide and a light outlet port for illuminating the area to be observed, the optical system preferably comprising an aspherical lens and a deflecting prism between the inlet port and the outlet port.

[0204] This configuration of the illumination system allows to obtain a well-collimated light beam at a working distance of preferably about 30 cm.

[0205] The choice of a liquid guide as light guide and a core diameter preferably of 5 mm or less allows reducing the divergence of the illumination beam and its size and increasing the intensity of the light in the center of the area to be observed (i.e., the cervix), thereby reducing parasitic reflections on surrounding surfaces (vaginal walls, speculum, etc.) and ultimately significantly improving the quality of the acquired images. The liquid light guide can provide illumination with satisfactory homogeneity, which is significantly higher than that obtained with quartz fiber bundles commonly used in colposcopy illumination systems.

[0206] Preferably, the inner diameter of the liquid light guide is in the range between 2.5 and 3.5 mm.

[0207] The lighting system may comprise an adapter configured to receive the light guide and to be mounted on the light source, possibly adjustable in three directions X, Y and Z relative to the light source.

[0208] Ergonomic colposcopy system

[0209] According to another aspect of the present invention, independently or in combination with the above, another object of the present invention is a colposcopy system comprising:

[0210] - Base with wheels;

[0211] - a column supported by a base with wheels;

[0212] - at least one reference reflector supported by a column for calibrating the polarimetry system;

[0213] -The vaginal lens head, supported by an articulated arm connected to the base.

[0214] A colposcopy system may also include:

[0215] - a workstation including a computer, supported by a base with wheels;

[0216] - a keyboard, supported by an articulated arm connected to the column; and

[0217] -The screen, supported by an articulated arm connected to the column, is located at a height higher than the height of the arm supporting the keyboard.

[0218] This arrangement allows the screen and keyboard to remain close to the user so that the user can easily control the computer themselves to acquire images.

[0219] The mobility of the system is also improved, as the assembly can be easily moved continuously over the ground.

[0220] Preferably, the system comprises a housing housing one or more light sources placed on a workstation, and a controller for controlling the various electro-optical and electronic components of the polarimetry system (such as controllable liquid crystal filters, filter wheels) and for processing images from the camera (if applicable) or even for generating polarimetry images, etc.

[0221] A reference reflector can be positioned within a light shield at the top of the column. This reflector is, for example, hinged about a vertical axis and comprises a frosted metal wall, for example made of aluminum, on one side and a reference surface, for example, with known spectral properties, on the opposite side. The rotatable mounting of the reflector allows one of the faces to be easily replaced with another without moving the head of the colposcope, which facilitates calibration operations. During these calibration operations, the head is positioned, for example, approximately 30 cm from the reflector by manipulating the articulated arm supporting it.

[0222] The colposcopic system advantageously comprises a pedal allowing the user to trigger a predefined action, eg start a polarimetric acquisition.

[0223] The colposcopic lens head may also include a button for triggering another predefined action (e.g., starting and stopping video recording). BRIEF DESCRIPTION OF THE DRAWINGS

[0224] The present invention will be better understood by reading the following detailed description of non-limiting embodiments of its various aspects and by referring to the accompanying drawings, in which:

[0225] Figure 1 is a block diagram of an example of a polarimetric multispectral imaging system according to the present invention;

[0226] Figure 2 is with Figure 1 Similar views illustrating the possibility of providing images that facilitate diagnosis;

[0227] Figure 3 The possibilities of using artificial intelligence to generate diagnostic aids are illustrated;

[0228] Figure 4 Schematic diagram of various information display modes in the colposcopy system according to the present invention;

[0229] Figure 5 is a diagram illustrating the acquisition of a series of successive measurements;

[0230] Figure 6 The diagram shows Figure 5 Generate an artificial series from the series of examples;

[0231] Figure 7 Pictured Figure 6 artificial reordering of measurements within a series;

[0232] Figure 8 illustrates the spatial recalibration of regions of an image corresponding to successive measurements;

[0233] Figure 9 The application of this method to measurements performed as part of Mueller polarimetry is illustrated;

[0234] Figure 10 The diagram shows Figure 9 The example case of artificial measurement series generation;

[0235] Figure 11 Assumptions used to accelerate the calculation of parameters of interest are illustrated;

[0236] Figure 12 shows comparative images obtained by implementing the Lu-Chipman decomposition on the one hand and the additive decomposition for accelerating the calculation of polarimetric parameters on the other hand;

[0237] Figure 13 is a schematic diagram and partial view of a colposcopy system according to the present invention;

[0238] Figure 14 More specifically, the colposcopic lens head and polarimetry and acquisition system are shown;

[0239] Figure 15 More specifically, the bottom parts of the colposcopic system are shown;

[0240] Figure 16 The diagram shows details of the system near the head of the colposcope;

[0241] Figure 17 is a top view of the colposcopy system deployed near the examination table;

[0242] Figure 18 It is a partial elevation view of the colposcopy system;

[0243] Figure 19 The colposcopic lens head is shown with parts of the polarimetry system attached to its front side;

[0244] Figure 20 is an exploded view showing the various optical components and their supporting parts of the polarization state generator and analyzer;

[0245] Figure 21illustrates the assembly of parts of the optical element supporting the polarization state generator;

[0246] Figure 22 illustrates the assembly of parts supporting the optical elements of a polarization state analyzer;

[0247] Figure 23 illustrates an isolated view of one of the support parts for attaching a filter holder for calibration;

[0248] Figure 24 The figure shows the assembly of two support parts of the polarization state generator;

[0249] Figure 25 A front view showing one of the support parts of the generator and the optical element received in the housing of this part;

[0250] Figure 26 is a front view of the front of the colposcope head equipped with a polarimetry system;

[0251] Figure 27 is a partial and schematic cross-sectional view of a lighting system;

[0252] Figure 28 yes Figure 27 Schematic and partial perspective views of the lighting system, with axial cross-sections;

[0253] Figure 29 is a partial and schematic longitudinal cross-sectional view of the colposcopic lens head near the light guide connection;

[0254] Figure 30 is an exploded perspective view of components for mounting a light guide on the head of a colposcope;

[0255] Figure 31 illustrates the spatial distribution of the luminous intensity in the vicinity of the observed area as a function of the diameter of the active part of the light guide used;

[0256] Figure 32 The calibration reflector and its shield are shown;

[0257] Figure 33 A standalone view of the reflector is shown, without the shield;

[0258] Figure 34 a calibration filter holder is shown schematically and partially;

[0259] Figure 35 shows a portion of a polarimetry system attached to the front of the colposcope head, provided with a protective cover;

[0260] Figure 36is a schematic and partial side view illustrating the installation of the acquisition system on the head of the colposcopic lens;

[0261] Figure 37 The filter wheel of the acquisition system and its drive motor are shown;

[0262] Figure 38 The acquisition system is partially and schematically shown in another perspective; and

[0263] Figure 39 An example of a multi-sensor camera optical system is shown schematically and partially. DETAILED DESCRIPTION

[0264] Multispectral polarimetry and / or color imaging systems

[0265] Figure 1 A multispectral polarimetric imaging system 1 according to the present invention is shown, for example, as a Mueller polarimetric colposcope, but it will be appreciated that this aspect of the invention is not limited to colposcopy.

[0266] The system 1 comprises an illumination optical system 2 for illuminating a target T (eg, the cervix), and an optical system 3 for directly observing the target and / or reconstructing an image of the target T on an electronic medium.

[0267] Illumination and observation of the target T takes place by means of a polarimetric system 4 comprising a polarization state generator (PSG) traversed by the light originating from the illumination system and a polarization state analyzer (PSA) traversed by the light originating from the target T.

[0268] Preferably, the PSG and PSA include a liquid crystal polarization modulator, particularly one whose control frequency is at least 60 Hz. Ferroelectric liquid crystal polarization modulators are preferred. System 1 also includes an acquisition system comprising a multi-sensor camera 5, for example, consisting of three CCD or three CMOS cameras, each with a sensor dedicated to the red, green, and blue regions of the visible spectrum.

[0269] The system 1 comprises a computer component 6 used, in particular, to manage the acquisition of intensity images for each sensor, control the operation of the liquid crystal polarization modulator, synchronize the acquisition of the camera with the modulation of the liquid crystal, process the images obtained by each sensor of the camera 5 and calibrate the system for each selected spectral band, as will be described below.

[0270] These computer components 6 include, for example, one or more processors, one or more microcontrollers, dedicated circuits (such as FPGAs or microcomputers), and associated human-machine hardware interfaces, and can be programmed to acquire intensity images in various spectral bands of interest (e.g., the red, green, and blue bands in the example under consideration), then calculate the Mueller matrix for these various spectral bands for all or some of the pixels of the image, and calculate and display the polarimetric parameters of interest. The calculations in the processing and post-processing steps for the Mueller matrix and various types of image processing operations can be performed very quickly using one or more graphics cards (GPUs). Various types of programming languages ​​(C, C++, Python, etc.) can be used for the calculations and for processing the images, as well as for managing the acquisition of the images.

[0271] Various types of programming languages ​​(C, C++, Python, etc.) can be used to calculate and process images, as well as manage image acquisition.

[0272] Using the same programming language used for acquiring images and for the processing and post-processing stages of the images can allow for completely eliminating the delay between the acquisition of an image and its real-time rendering.

[0273] Thus, the computer means 6 may comprise a controller for controlling the PSG and the PSA and for processing images originating from the camera, said controller comprising, for example, one or more FPGAs, and a workstation comprising a computer, equipped, where applicable, with at least one graphics card, and a human-machine interface comprising, for example, a screen, a keyboard and one or more control buttons or pedals, as described below.

[0274] The illumination system is capable of emitting in each of the observation spectral bands.

[0275] In the case where the spectral bands are in the red, green and blue bands respectively, the lighting system may include a white light source (such as a xenon source associated with a spectral filter), or a group of LED diodes emitting in the red, green and blue bands respectively, or LED diodes emitting in the blue and yellow bands.

[0276] Use of a xenon source is preferred because of the ease of use it offers and due to its power, and because it is already widely used in endoscopy systems.

[0277] Since the camera of the acquisition system is a multi-sensor camera, preferably a triple CCD or triple CMOS RGB camera, it can acquire several images simultaneously without significant light losses and without any interference problems between channels ("crosstalk").

[0278] The image obtained on each sensor may be a grayscale intensity image for the corresponding spectral band.

[0279] The grayscale intensity images of the various sensors can be perfectly superimposed by construction, thereby simplifying the reconstruction of the final image after processing, in particular adding polarimetric information to a color image or combining polarimetric images of various wavelengths.

[0280] The images obtained by the multispectral polarimetry imaging system applied to Mueller polarimetry in various spectral bands (especially the red R, green G and blue B bands) can be generated by performing calculations in various spectral bands, such as Figure 1 If applicable, these images can be combined to generate a Mueller polarimetry RGB color image.

[0281] At the same time, the intensity images acquired by the camera's three sensors can be combined to form an RGB color image. This image is useful for allowing practitioners to clearly identify the area where polarimetric analysis is to be performed.

[0282] Determining the Mueller matrix for each of the R, G, and B spectral bands allows calculation of intensity images in each of these bands that express the polarimetry parameter value (e.g., hysteresis) for each pixel of the image, as Figure 2 As shown in .

[0283] Thus, for example, a depolarized image is generated for each spectral band, which image is determined by knowing the Mueller matrix, and a delayed image is generated, which image is also determined by knowing the Mueller matrix.

[0284] One or more lag and / or depolarized images may be combined with the color image to generate an image I in which the hue and / or contrast of some areas are modified to provide additional information to the practitioner to aid in diagnosis.

[0285] Combining a liquid crystal polarization modulator polarimeter, a multispectral illumination source (especially a xenon lamp) and a triple CCD or triple CMOS camera in the same imaging system allows to obtain a particularly compact and efficient imaging system that is especially suitable for colposcopy.

[0286] The fact that it is possible to generate multimodal images that are perfectly spatially calibrated with respect to each other facilitates deep learning, e.g. with the help of convolutional neural networks7, e.g. Figure 3 As shown in , this network receives as input a conventional color image, a monochrome intensity image, and a polarimetric image for a selected wavelength range and outputs one or more simplified images for aiding diagnosis, e.g., in the form of enhanced contrast.

[0287] In particular, during learning (which can be supervised learning), it is possible to provide the artificial intelligence system with non-polarimetric RGB color images, as well as polarimetric images representing images related to, for example, depolarization, linear phase lag, and the azimuth of the linear phase lag, wherein the artificial intelligence system generates one or more images containing information that assists in diagnosis.

[0288] The computation component 6 may be configured to provide all or some of the following visualizations:

[0289] - Parallel multimodal visualization of various polarimetric and non-polarimetric images ( Figure 4 A) For example, an RGB color image (left), a linear phase lag image (middle), and an azimuth image of the slow axis (right) obtained by a camera are displayed on the same screen;

[0290] -Multi-spectral parallel visualization ( Figure 4 B), for example, a reference RGB color image on the left and a series of three triple images in each spectral band of interest on the right, the first row representing linear phase lag, the second row representing the azimuth of the slow axis of the linear phase lag, and the third row representing depolarization;

[0291] -Interactive visualization ( Figure 4 C), where the overall color image RGB is displayed, but the display area is limited, in this case a circle, e.g. centered around a pointer that the user can move on the image, revealing information related to a polarimetric parameter, in this case the linear phase lag, partially superimposed on the displayed RGB image;

[0292] -Visualization( Figure 4 D) The left column shows an RGB color image (top), a color polarimetry image (middle), and an image of the azimuth of the slow axis of the linear phase lag at the bottom, as well as a magnified image of the three images in the left column merged together (right).

[0293] It may also display:

[0294] - Color image of the Mueller matrix;

[0295] - depolarized color image;

[0296] - Color image of linear phase lag;

[0297] - Color images of the azimuth of the linear phase lag, etc.

[0298] Such an imaging system is advantageously applied in colposcopy, as will be described below, but may also be applied in other types of imaging systems, such as endoscopy, and microscopy and exoscopy for neurosurgery.

[0299] A review of Mueller polarimetry

[0300] Generally speaking, the polarization state S of the incident light in The light S that excites the sample out The following relationship exists:

[0301] S out =M*S in

[0302] where M is the Mueller matrix.

[0303] In order to obtain the 4×4 Mueller matrix M, the intensity coefficients need to be measured. These intensity coefficients can be obtained by using a polarization state generator (PSG), which generates a polarization state generator consisting of four Stokes vectors Four independent polarization states are characterized.

[0304] Each represents a column of the modulation matrix W:

[0305]

[0306] After interacting with the observed area, each polarization state generated by the polarization state generator is analyzed by four polarization configurations of the polarization state analyzer (PSA).

[0307] These four configurations are also represented by the four Stokes vectors representing the rows of the analysis matrix A describe:

[0308]

[0309] After these steps the following intensity matrix B is obtained:

[0310] B=AMW where:

[0311] M=A -1 BW -1

[0312] B corresponds to a series of n measurements, in this case 16 measurements in the case of a 4×4 Mueller matrix.

[0313] According to this model, the measured intensity matrix B can be written as the sum of the polarization contribution and the non-depolarization contribution.

[0314] B=B nd +B d =A(qM nd +pM d )W

[0315] M=qM nd +pM d =A-1 (B nd +B d )W -1

[0316] Various known methods can be used to process the Mueller matrix in order to calculate the polarimetric properties, i.e., depolarization, linear phase lag, etc. In the case of a polarimetric imaging system according to the present invention, 16 successive measurements can be made for each pixel of the image for each sensor of a multi-sensor camera, corresponding to the respective coefficients of the intensity matrix B.

[0317] Preferably, the Mueller matrix is ​​a regular 4x4 matrix, i.e. a complete Mueller matrix, but using an incomplete 3x3, 3x4 version or another version does not depart from the scope of the invention. The invention can also be used for simplified polarimetric imaging techniques such as Stokes polarimetric imaging.

[0318] Increase the frequency of images

[0319] According to one of its aspects, the present invention allows for an increased flow of results based on a continuous series μ of n actual measurements per unit time of a physical quantity (or set of physical quantities) X associated with a physical object O (referred to as the original series). These measurements are performed sequentially in a predefined order within each given time interval, while generating additional series of the original series (referred to as artificial series). In the case of Mueller polarimetry, these measurements can be 16 intensity measurements of a matrix B, but this aspect of the invention is more general and can be applied to other types of imaging than Mueller polarimetric imaging. Thus, X can be a matrix other than the Mueller matrix obtained by the Mueller polarimetric measurement system, in particular a non-square real matrix, a square real matrix, a vector, or even a real number, or more generally any physical property or set of physical properties whose determination involves several consecutive measurements. Alternatively, X can be obtained for an array of points (imaging). This table can be two-dimensional (2D) or three-dimensional (3D). This is the case for 2D or 3D imaging. For 2D imaging, each measurement point corresponds to a pixel. For 3D imaging, each measurement point corresponds to a voxel. In order to correctly calculate the result X=G(μ), the series of n measurements of μ must follow a predefined order: μ0,μ1,μ2,…μ n-1 .

[0320] Any other order, such as μ'=(μ2, μ3, ..., μ n-1 ,μ0), generally produces the value X'=G(μ'), where X' is not the correct result.

[0321] If the measuring device can perform l times (l∈N) in 1 second, μ=(μ0, ..., μn-1 ), then it provides a total of s discrete measurement sets m = (m0, ..., m s-1 ), where s = ln.

[0322] These measured values ​​are, for example, the coefficients of the intensity matrix B in the case of Müller polarimetry.

[0323] Figure 5 Three actual series μ of measurements are shown as examples, each series comprising four measured values ​​(n=4). μ0 represents the first value, μ1 represents the second value, μ2 represents the third value, and μ3 represents the last value, this order being predefined.

[0324] First, the estimated result X (e.g., the Mueller matrix) does not change or hardly changes with time. Therefore, it can be assumed that each series μ has n measurement values ​​that are similar to the corresponding values ​​of the next series obtained in the same order.

[0325] It is then possible to artificially form new combinations of measurements by correlating measurements from different series, for obtaining new series of artificial measurements, e.g. Figure 6 As shown in .

[0326] For example, in Figure 6 In the example, for the series The values ​​μ1, μ2, μ3 of the first series are associated with the value μ0 of the second series.

[0327] These groups are repeated on all n measurements.

[0328] series and and Figure 5 The series μ of measurements remain the same. The order of measurements remains unchanged. There is no need to permute these series of measurements.

[0329] For the series and The predefined order of measurements is not followed; a permutation function P is then applied, which Figure 7 The diagrams are arranged in such a way that a predefined order of measurements is followed in each series.

[0330] In the series After the permutation, in order to maintain the predefined order, the function G can be applied to all new permuted series In order to calculate X.

[0331] In the general case of n measurements, if T n is the time required to make n measurements, T tis the time required to make one measurement, then 1 / T of n measurements can be made in one second n series; T p is the time required to perform one substitution, and T c is the time required to calculate X=G(μ), so preferably T c < <T t And T p < <T t .

[0332] As indicated previously, the measurement results may be successive intensity values ​​for each pixel of an image recorded by a given sensor of a camera of the acquisition system, and the above method may be applied to each pixel of this image.

[0333] However, this image may change over time, for example due to movement of the object.

[0334] Figure 8 (Left) The simplified case described previously for n = 4 is considered again. Each small square represents a portion of the image in the entire field of view represented by the large square. This portion of the image can move within the field of view over time.

[0335] It is then worthwhile to apply a spatial recalibration function R to recalibrate the images acquired at different successive times, as shown (right), so that the successive intensity measurements used in calculating the Mueller matrix actually relate to the same region.

[0336] The recalibration function R can be obtained (for example, from Figure 8 A series of measurements of the image corresponding to m4 in the acquisition) is applied immediately after the new image.

[0337] The recalibration allows the pixels corresponding to values ​​m1, m2, and m3 to coincide spatially with the pixel corresponding to m4, the pixels corresponding to values ​​m2, m3, and m4 to coincide spatially with the pixel corresponding to m5, the pixels corresponding to values ​​m3, m4, and m5 to coincide spatially with the pixel corresponding to m6, and so on. Figure 8 As shown in .

[0338] The image recalibration function may involve any suitable image recalibration algorithm; an example of a recalibration function is described in M. Irani and SP Peleg, 1991, in CVGIP Graphical models and image processing, Elsevier, entitled "Improving Resolution by Image Registration".

[0339] The recalibration function may seek to determine a recalibrated value a in x and a recalibrated value b in y between two images that minimizes a loss function between them.

[0340] Examples of loss functions are provided in M.B.A. Haghighat, A. Aghagolzadeth, and H. Seyedarabi, “A non-reference image fusion metric based on mutual information of image features,” Computers & Electrical Engineering, Vol. 37, No. 5, pp. 744-756, September 2011, doi:10.1016 / j.compeleceng.2011.07.012.

[0341] The permutation function P can be applied after the recalibration function R to put the measurements in the correct order, and then the function G can be applied to calculate the result X.

[0342] When seeking to calculate the result X more than eight times per second, in order to have a degree of smoothness corresponding to essentially real-time imaging, 1 / Tn>=8.

[0343] The process described above can also be used in the more general case where the object O is deformable. If Td is the characteristic time of the deformation, then preferably Td>>Tn.

[0344] If Tx is a characteristic duration of a change in a property exhibited by outcome X, then preferably Tx>>Tn.

[0345] In the case of Mueller polarimetry imaging, the measured values ​​are the coefficients B of the intensity matrix B ij The value of (i,j=1,…,4).

[0346] The sixteen coefficients of this matrix correspond to the measurements made for a given pixel, so that the Mueller matrix for this pixel can then be calculated and the series of measurements μ formed, such as Figure 9 As shown in .

[0347] Figure 10 The diagram shows the permutation function P(μ) for the series application.

[0348] In this case, the function G corresponds to the calculation allowing the Mueller matrix M to be obtained from the intensity matrix B.

[0349] M=G(μ)=X=A -1 BW -1

[0350] The recalibration function R is applied before applying the permutation function P; the recalibration function R may be applied by taking as reference image, for example, the image corresponding to the first image in a sequence of 16 consecutive images (in the case of a 4x4 Mueller matrix).

[0351] Preferably, T p +T R +T c <T t , in order to have real-time imaging, where:

[0352] T p is the duration required to displace P;

[0353] T R is the duration required to apply the recalibration function R;

[0354] T c is the duration required to calculate M;

[0355] T t is the duration of a single measurement.

[0356] In the case of a portion of the spectral range corresponding to the blue, green and red parts of the visible spectrum being acquired simultaneously, the coefficients of the unnormalized and unpolarized intensity of the Mueller matrix obtained in the blue, green and red parts of the visible spectrum respectively using the above-mentioned method are can be combined to produce a color image of the target in real time.

[0357] Mueller matrix post-processing

[0358] Conventional calculation of the Mueller matrix can involve, for example, a decomposition known as the Lu-Chipman decomposition, as described in an article by S.-Y. Lu and RA Chipman entitled “Interpretation of Mueller matrices based on polar decomposition,” published in Journal of the Optical Society of America A, Vol. 13, No. 5, p. 1106, May 1996, doi: 10.1364 / JOSAA13001106.

[0359] However, this decomposition is relatively demanding in terms of computational time since it involves computing the eigenvalues ​​of the matrix M.

[0360] Furthermore, when observing the cervix, mainly birefringence and depolarization effects were observed.

[0361] In this case, it is assumed that the non-depolarizing effect is associated with the surface area of ​​the observed region, while the depolarizing effect of its part is related to the volume of the observed region, e.g. Figure 11 As shown in .

[0362] This figure schematically illustrates the propagation of light within tissue.

[0363] In this case, the Stokes vector S nd is associated with the non-depolarized component originating from the surface layer of the tissue, while the Stokes vector S d Associated with the depolarized component of light originating from the volume of tissue.

[0364] Figure 11 Also shown are the two Mueller matrices M associated with the non-depolarized and depolarized light components, respectively. nd and M d .

[0365] The factors q and p are intended to estimate the ratio of non-depolarized light to depolarized light and must be determined.

[0366] The intensity matrix B can be formulated as the sum of the contributions of the non-depolarized light and the polarized light.

[0367] then:

[0368] B=B nd +B d =A(qM nd +PM d )W

[0369] and

[0370] M=qM nd +pM d =A -1 (B nd +B d )W -1

[0371] Non-depolarized light M nd The Mueller matrix of can be corresponded to the linear lag Mueller matrix, such as:

[0372] θ: azimuthal orientation of the fast axis, in degrees

[0373] δ: Phase lag, in degrees

[0374]

[0375] Mueller matrix M of depolarized light d can be equal to:

[0376]

[0377] then

[0378]

[0379] The Mueller matrix M can be expressed relative to its unnormalized and unpolarized intensity coefficients Normalized and can be written as:

[0380]

[0381] in

[0382]

[0383] The matrix Mnormalised can be written as:

[0384]

[0385] The coefficient m ij (i, j = 1, 2, 3, 4) is the normalized matrix M relative to its unnormalized and unpolarized intensity coefficients The coefficient of .

[0386] Depolarization can be calculated using the following formula:

[0387]

[0388] This decomposition method provides significant time savings since time-consuming calculations of eigenvalues ​​and eigenvectors are not required.

[0389] Moreover, this approach is matrix-based, which allows the algorithm to perform multiple calculations simultaneously rather than serially, which saves a lot of time and memory.

[0390] Various polarimetric parameters can thus be obtained more quickly, which is very advantageous in the context of real-time polarimetric imaging.

[0391] Figure 12 Shown is a comparison of polarimetric images obtained via conventional Lu-Chipman decomposition on the one hand and via the additive decomposition described above on the other hand.

[0392] The top three images are obtained by additive decomposition as described above and correspond to the display of linear phase lag δ (left image), azimuthal angle of linear phase lag θ (middle image), and depolarization (right image), respectively, while the bottom three images represent the same parameters obtained via Lu-Chipman decomposition.

[0393] Significant similarities can be seen between images representing the same parameters, which demonstrates the performance of the method.

[0394] Colposcopy system

[0395] The imaging system according to the present invention may be designed to perform colposcopy, for example, using Figure 13 The colposcopic system 10 is shown in FIG.

[0396] The system 10 comprises a colposcope 20 having a head 21 supported by an articulated arm 22 which is itself supported by a base 31 on wheels.

[0397] The vertical column 30 is also supported by a base 31 on wheels.

[0398] The colposcope 20 is connected to a lighting system comprising a light source 50 housed in a housing connected to the head 21 by a flexible light guide 51. Figure 13 Not visible in but Figure 14 As can be seen in FIG. 5 , the housing housing the light source 50 rests on a workstation 60 comprising a computer, for example.

[0399] The colposcopic system 10 comprises a screen 40 connected to a computer 60, the screen advantageously being a touch screen, preferably liquid-proof.

[0400] This screen is supported by a column 30 and is connected to a computer 60 as is the keyboard 41 .

[0401] A controller 70 is connected to the computer 60 and is used in particular to control the operation of the polarimetry system 90 supported by the head 21. This controller 70 may also be connected to the acquisition system 110, and in particular to the camera 5.

[0402] Preferably, if Figure 32 and Figure 33 As seen more specifically in FIG, the colposcopy system 10 comprises at least one reflector 81 housed in a support supported by the column 30 and used to calibrate the polarimetry system. The reflector is covered by a light shield 80 which is preferably supported by the column 30, in particular at its top.

[0403] This reflector 81 is, for example, hinged about a vertical rotation axis and comprises, for example, a frosted metal wall 82 (e.g. made of aluminum) on one side and a reference surface 83 with known spectral properties (e.g. a substrate with a neutral color surface with high Lambertian reflectivity, called a "spectralon") on the opposite side.

[0404] The swivel mounting of the reflector 81 allows one of the faces to be easily replaced by another without moving the colposcope's head 21, which facilitates calibration operations. During this calibration, the head 21 is placed, for example, about 30 cm from the reflector 81 by manipulating the articulated arm supporting the head.

[0405] The colposcopic system 10 advantageously includes a pedal 100, such as Figure 13 As shown in FIG, the pedal 100 allows a user to trigger a predefined action, for example, to start a polarimetry acquisition.

[0406] The vaginal lens head 21 may further include a button 101. Figure 16 , used to trigger another predefined action, for example, starting and stopping video recording.

[0407] Figure 19 The polarization measurement system 90 shown in FIG. 1 includes a polarization state generator PSG and a polarization state analyzer PSA, and almost all of its optical components are in Figure 20 Shown in.

[0408] The polarization state generator PSG comprises a series of four optical elements placed in the path of the light exiting through the window 210 of the head 21 of the colposcope, namely, in a direction extending from the outside towards the exit window 210, a first 510 nm half-wave liquid crystal polarization modulator HFLC 191, a 633 nm quarter-wave plate QWP 192, a second 510 nm quarter-wave liquid crystal polarization modulator QFLC 193 and a fixed linear polarizer 194.

[0409] The polarization state analyzer PSA comprises a series of three optical elements placed in the path of light arriving at one of the entrance ports 211a and 211b associated with the left and right eyepieces respectively (in this case, the entrance port 211b associated with the practitioner's right eyepiece).

[0410] Starting from the outside and towards the head of the colposcope, there are: a first 510 nm half-wave liquid crystal polarization modulator HFLC 221 , a 633 nm quarter-wave plate QWP 222 and a second 510 nm quarter-wave liquid crystal polarization modulator QFLC 223 .

[0411] The analyzer also includes a linear polarizer filter 224 similar to the polarizer 194 of the PSG, disposed downstream of the camera 5 within the acquisition system 110, as shown in FIG. Figure 36 As shown in .

[0412] In the example shown, each optical component 191 , 192 or 193 of the PSG is circular in shape and is held between two mutually attached support parts, one of which has a housing 233 for receiving the element and the other of which serves to hold the element in its housing.

[0413] All of these support parts have a C-shaped portion that opens upward and leaves the upper edge of each optical element clear.

[0414] The polarizer 194 has a square or rectangular outline and is housed in a support member 265 provided with a slide member that opens upward.

[0415] The optical element 191 is placed between a first support part 261 and a second support part 262 , which has a housing 233 that receives the element 191 , wherein the support part 261 forms a retaining cover.

[0416] The support part 262 serves as a cover for a third support part 263 having a housing 233 for receiving the optical element 192, such as Figure 21 As shown in .

[0417] This third support part 263 serves as a cover for a fourth support part 264 which has a housing 233 which receives the optical element 193 .

[0418] The various support parts can be overlapped and attached to the support 265 by screws 242 through the base of the PSA in front of the head 21 of the colposcope, as shown in FIG. Figure 19 As shown in .

[0419] The optical elements 221 , 222 and 223 of the PSA are also circular in shape and are supported by support members 251 , 252 , 253 and 254 that overlap one another.

[0420] Each support element 251 to 254 has a base 255 provided with holes for the screws 242 for attachment to the front face of the head 21, and a generally C-shaped upper portion 256 oriented substantially downward at 45° and connected to the base 255 by a post 257, as shown. Figure 22 As shown in .

[0421] The support parts 252 to 254 each have a housing 258 that accommodates a corresponding optical element.

[0422] Figure 22 The optical element 221 is shown held between support parts 251 and 252, support part 251 acting as a cover, and the optical element 222 is held between support parts 252 and 253 acting as a cover, and element 223 is held between support parts 253 and 254 acting as a cover.

[0423] The upper portion of the support part 252 supports two branches 270 for attaching a filter holder 310, such as Figure 34In the example considered, this filter holder 310 supports three filters 311, 312 and 313, which are respectively a phase retarder L30 whose fast axis is oriented at 30° relative to the linear polarizer P0, a linear polarizer P90 whose transmission axis is oriented at 90° relative to the polarizer P0 and a polarizer P0 whose transmission axis is oriented at 0° relative to the reference axis.

[0424] The filter holder 310 also includes a clear area that is placed in front of the PSA after calibration is complete to allow acquisition of an image.

[0425] In the example considered, the filter holder 310 is laterally movable in a substantially horizontal direction.

[0426] Electronic circuit 320 allows computer 60 to know the position of the filter holder and therefore which filter is active (if applicable).

[0427] During calibration, the filter holder 310 may be motorized or may be manually moved to sequentially place the individual filters 311 to 313 in front of the PSA.

[0428] Calibration can be performed according to a method known as the Eigenvalue Calibration Method (ECM).

[0429] Figure 25 In particular, the housing 233 is shown to be laterally open to allow control wires for the optical elements received therein to pass through the channel 233a.

[0430] The housing 233 also has a gap 233b opposite the channel 233a for receiving the control wire near its connection with the optical element. The control wire follows the contour of the optical element in a semicircular channel 233c that conforms to the outer shape of the optical element.

[0431] Figure 23 Optical elements 221 and 223 of the PSA having larger diameters are shown to almost completely overlap at the two inlet ports 211a and 211b, while optical element 222 having a smaller diameter than elements 221 and 223 completely overlaps port 221b but partially overlaps port 211a.

[0432] However, given the focusing distance, this in no way affects the quality of the observed image, since focusing occurs at a much greater distance than the distance of the optical element from port 221a.

[0433] Figure 27 is a partial and schematic cross-section of the light emitting source 50 .

[0434] This light source includes a xenon lamp 52 and a set of filters 53 and 54, namely, a first bandpass filter (Edmund Optics #84-728) that allows only light between 400 and 750 nm to pass through with a transmittance of approximately 95% and a second high-pass filter (Edmund Optics #84-754) that allows only light with a wavelength greater than 400 nm to pass through with a transmittance of approximately 93%. The selection of these filters allows a relatively high transmittance to be achieved, approximately 90% in the 400-750 nm range.

[0435] One end 51 a of the light guide 51 is held on the axis of the lamp 52 by means of an end piece 56 .

[0436] like Figure 29 As shown in FIG, the other end 51b of the light guide 51 is located on the axis of the aspherical lens 180 of the colposcope, thereby providing an image of the liquid guide at a distance of about 30 cm from the colposcope head.

[0437] The lens 180 is placed in front of a prism 181 which returns the light towards the exit window of the colposcope.

[0438] If applicable, the light guide 51 can be held by means of a set of supports 191 and 192 whose positions relative to the body of the head of the colposcope can be adjusted along X and Y respectively to allow precise adjustment of the position of the end 51b on the axis of the lens 180.

[0439] It is worthwhile to use a liquid guide 51 whose core has a diameter of 3 mm, because this means that, as Figure 31 As shown in , it is possible to benefit from a stronger illumination in the center of the area illuminated by the colposcope compared to a conventional guide whose core has a diameter of 5 mm. In particular, a liquid guide with a diameter of 3 mm allows obtaining a higher illumination on a surface with a diameter of 3 cm, which corresponds approximately to the diameter of the cervix, than is obtained with a liquid guide with a diameter of 5 mm.

[0440] like Figures 36 to 38 As can be clearly seen in the figure, the acquisition system 110 is attached to the head 21 of the colposcope and receives the light reaching one of the eyepieces (the same eyepiece in front of which the PSA is placed) through one or more deflection prisms integrated in the head 21.

[0441] In the example considered, the camera 5 is equipped with an objective lens 415 and a filter wheel 411 supporting three filters 421 , 422 and 423 .

[0442] For example, it is a three-band filter suitable for the camera 5 , a bandpass filter centered at 650 nm and having, for example, a spectral width of 40 nm, and a bandpass filter centered at 700 nm and having a spectral width of 50 nm.

[0443] The filter wheel 411 is arranged to be rotated by a motor 410. A system of contactor angle markings may allow the system to know the angular position of the wheel and therefore the angular position where exactly the filter is positioned on the path of light reaching the camera 5.

[0444] like Figure 39 As shown in , camera 5 may include three dichroic prisms for separating wavelengths toward the various sensors. For example, the camera includes dichroic prisms 605, 604, and 606, where blue detection sensor 603 is attached to prism 605, red detection sensor 601 is attached to prism 604, which is itself attached to prism 605, and green detection sensor 602 is attached to prism 606, which is itself attached to prism 604.

[0445] To use the system 10 , the user may perform polarimetric calibration using the two faces of the reflector 81 and the various filters of the filter holder 310 in a known manner.

[0446] The head 21 can then be positioned to image the cervix after the vagina has been dilated using the speculum.

[0447] At any time, the practitioner can initiate video recording by pressing button 300 and / or can trigger polarimetric acquisition by pressing pedal 100. Pressing the pedal triggers polarimetric acquisition by stopping video recording, which automatically resumes once the polarimetric acquisition is complete. Pressing the button again stops video recording.

[0448] Of course, the present invention is not limited to the examples described above.

[0449] The optical system of the colposcope can in particular be modified, for example by removing the eyepiece, so that observation can then only be performed on a screen.

Claims

1. A real-time imaging method comprising the following steps: - Obtain the original actual series μ of n measurements at given time intervals, μ = (μ0, ..., μ n-1 ), these n measurements succeed one another in a series in a predefined order, and by processing these measurements, each series of n measurements allows generating at least one raw result X=G(μ) at a given frequency f; - generating an artificial series from a set of measurements originating from a real series μ with a given rank and from at least one real series with a different rank, in particular the next rank, and from a permutation P of the measurements within each artificial series so as to observe a predefined order of the n measurements within each series; - Generate a stream of results with a frequency greater than f from the original results and from results generated from artificial series that are between the real series.

2. The method of claim 1, applied to polarimetric imaging.

3. The method of claim 2, applied to colposcopy, microscopy, encephaloscopy, endoscopy, preferably to polarimetric colposcopy, and more preferably to multispectral polarimetric colposcopy.

4. The method as claimed in claim 1, wherein the result X is a Mueller matrix (M) or a physical property calculated from this matrix, and wherein the series of measurements are polarimetric measurements.

5. Method according to claim 4, said series of measurements corresponding to the acquisition of coefficients of an intensity matrix B, preferably at least 9 coefficients of the intensity matrix B, in particular 12 coefficients, more preferably 16 coefficients.

6. The method according to claim 5, wherein the Mueller matrix is ​​a 4×4 matrix, wherein B=AMW, in as well as The measurement is generated using four Stokes vectors The four independent states are described by the polarization state generator (PSG) and generated by the four Stokes vectors The analysis states described are obtained with a polarization state analyzer (PSA).

7. The method of claim 6, wherein the additive decomposition of the Mueller matrix M is M = qM nd +pM d The form of execution, where M nd is the non-depolarized component and M d is the depolarization component, and the parameters q and p are the weights of the two components of the Mueller matrix, M nd In particular, the Mueller matrix of a linear phase lag is given by: And M d In particular, the Mueller matrix of a pure depolarizer is given by: where θ is the azimuthal orientation of the fast axis in degrees, and δ is the phase lag in degrees (ranging from 0° to 180°); The coefficient m ij (i, j = 1, 2, 3, 4) are the coefficients of the normalized Mueller matrix, is the unnormalized and unpolarized intensity coefficient of M, and the other coefficients m ij (i, j = 1, 2, 3, 4) relative to is normalized.

8. The method of claim 7, wherein depolarization is provided using the following formula: 9 . The method according to claim 7 , wherein the Mueller matrix and the parameters are obtained for a single wavelength or simultaneously for several wavelength ranges, in particular at least two wavelength ranges, in particular wavelength ranges in the visible and infrared spectral ranges.

10. The method according to any one of claims 7 to 9, applied to coefficients of the unnormalized and unpolarized intensities from three Mueller matrices obtained simultaneously in the spectral ranges corresponding respectively to the blue, green and red parts of the visible spectrum to reconstruct the color image in real time.

11. The method according to any one of the preceding claims, wherein Before performing the permutation P, a spatial recalibration is performed on the images corresponding to the measurements in order to take into account possible displacements of the observed area in the field of view of the camera.

12. A multispectral polarimetric imaging system (10), in particular for colposcopy, comprising: - a lighting system comprising at least one light emitting source (50), the lighting system emitting in at least two, in particular at least three, spectral bands; - a polarization state generator (PSG) disposed downstream of the light source and upstream of the target (T) to be imaged; - A polarization state analyzer (PSA), deployed downstream of the target to be imaged; - a multi-sensor camera (5) comprising at least two, more preferably at least three, sensors (601, 602, 603) for recording at least two, more preferably at least three, images in said spectral bands, respectively.

13. The system of claim 12, involving three spectral bands, and preferably, the first spectral band ranges from 445nm to 475nm, the second spectral band ranges from 510nm to 550nm, and the third spectral band ranges from 600nm to 660nm, these spectral bands are preferably centered around 460nm, 530nm and 630nm respectively.

14. The system of claim 12 , wherein the camera ( 5 ) is a multi-CCD camera or a multi-CMOS camera, in particular a dual-CCD camera, a dual-CMOS camera, a triple-CCD camera, a triple-CMOS camera, a 4CMOS camera, a 4CCD camera, more preferably a triple-CCD camera or a triple-CMOS camera, preferably a triple-CMOS camera.

15. The system of claim 11 involving three spectral bands and located in the red / near infrared, green and blue bands respectively.

16. A system as claimed in any one of claims 12 to 15, the camera comprising at least two, more preferably at least three, dichroic prisms for separating the wavelengths towards the respective sensors.

17. The system of any of claims 12 to 16, the polarization state generator (PSG) comprising an electrically controllable liquid crystal polarization modulator.

18. The system of any one of claims 12 to 17, the polarization state analyzer (PSA) comprising an electrically controllable liquid crystal polarization modulator.

19. The system of any one of claims 12 and 18, the liquid crystal polarization modulator being a ferroelectric or nematic liquid crystal polarization modulator, preferably a ferroelectric liquid crystal polarization modulator.

20. The system of any one of claims 12 to 19, the light source being a white light source, more preferably a xenon lamp.

21. A system as claimed in any one of claims 12 to 20, comprising a three-band dichroic filter downstream of the light emitting source.

22. The system of any one of claims 12 to 21, comprising a filter wheel comprising three filters (421, 422, 423) disposed in front of the camera.

23. A system as claimed in claim 22, the filter wheel supporting a multi-band filter, in particular a tri-band filter, preferably for allowing acquisition by the camera (5) of wavelengths in spectral bands centered around 460nm, 530nm and 630nm, and one or more monochromatic dichroic filters, preferably for allowing acquisition of images at 650nm and 700nm, respectively.

24. The system of any one of claims 12 to 23, comprising a processor for generating at least one Mueller polarimetric image of the target in each of the spectral bands.

25. The system of claim 24, the processor being configured to superimpose the image containing the at least one item of polarimetric information at least partially on an image corresponding to a non-polarimetric observation made by the camera.

26. The system according to claim 24 , wherein the system is designed to establish a real series of n measurements corresponding to at least 9 coefficients, in particular 12 coefficients, more preferably 16 coefficients, of the acquisition intensity matrix B, the processor being designed to: - generating a series of artificial measurements from a set of measurements originating from a real series μ of measurements with a given rank and from at least one real series with a different rank, in particular the next rank, and from a permutation P of the measurements within each artificial series so as to observe a predefined order of the n measurements within each series; - generating a stream of polarimetric images from polarimetric images generated by the actual series of measurements and from the artificial series between the actual series at a frequency greater than the frequency of the actual series that would allow said measurements but not the artificial series.

27. A system as claimed in claim 26, the processor being designed to perform a spatial recalibration of the images corresponding to the measurements before performing the permutation P, so as to take into account any possible displacement of the observed area within the field of view of the camera.

28. The system of any one of claims 24 to 27, wherein the processor is configured to perform a calculation on the Mueller matrix M of the form M = qM. nd +pM d Additive decomposition of M nd is the non-depolarized component and M d is the depolarization component, and the parameters q and p are the weights of the two components of the Mueller matrix, M nd In particular, the Mueller matrix of a linear phase lag is given by: And M d is the Mueller matrix of a pure depolarizer given by: where θ is the azimuthal orientation of the fast (or slow) axis in degrees, and δ is the phase lag in degrees (ranging from 0° to 180°); in The coefficient m ij (i, j = 1, 2, 3, 4) are the coefficients of the normalized Mueller matrix, is the unnormalized and unpolarized intensity coefficient of M, and the other coefficients m ij (i, j = 1, 2, 3, 4) relative to is normalized.

29. The system of claim 28, wherein the processor is configured to calculate depolarization by performing the following operations:

30. The system of any one of claims 12 to 29, being designed to generate a parallel display of at least one non-polarimetric image and at least one polarimetric image in at least one of the spectral bands.

31. The system of any one of claims 12 to 30, being a colposcopy system, comprising: - a base with wheels (31); - a column (30) supported by a base with wheels; - a colposcope head (21) supported by an articulated arm connected to a base; - at least one reference reflector (81) supported by the column (30) for calibrating the polarimetry system, and preferably also comprising: - a workstation including a computer (60) supported by a base with wheels; - a keyboard (41) supported by an articulated arm connected to the column; - A screen (40) supported by an articulated arm connected to a column at a height higher than that of said arm supporting the keyboard.

32. A system as claimed in claim 31, wherein the reference reflector (81) is deployed in a light shield (80) on top of the column, hinged about a vertical axis, comprising a frosted metal wall on one side, preferably made of aluminum, and a reference surface with known spectral characteristics on the opposite side.

33. A system as described in any one of claims 31 and 32, comprising a pedal (100) allowing a user to trigger a predefined action, in particular to start polarimetry acquisition, the head (21) of the colposcope comprising a button (101) for triggering another predefined action, in particular for starting and stopping video recording.

34. The system of any one of claims 12 to 33, being a colposcopy system, comprising an illumination system, said illumination system comprising: o at least one light source (50); a liquid light guide (51) connected at one end to a light emitting source, the light guide having a core diameter of 5 mm or less; The system comprises a colposcope head (21) comprising an optical system having a light inlet port connected to the other end of the liquid light guide and a light outlet port for illuminating the area to be observed.

35. The system of claim 34, the optical system comprising an aspheric lens (180) and a deflecting prism (181) between the light inlet port and the light outlet port.

36. A system as claimed in claim 34 or 35, the liquid light guide (51) having an inner diameter in the range between 2.5 and 3.5 mm.

37. A method for training an artificial intelligence system (7), preferably comprising at least one convolutional neural network, wherein the artificial intelligence system receives as input non-polarimetric images and polarimetric images generated by a polarimetric imaging system according to any one of claims 12 to 36.

38. An observation system with two inlet ports, in particular a binocular system, in particular for colposcopy, comprising a head (21), the head comprising an optical system having a light outlet port for illuminating the area to be examined and left and right inlet ports (221a, 221b) pointing towards the area to be observed, the observation system further comprising a polarisation measurement system (90), the polarisation measurement system (90) comprising a polarisation state generator (PSG) arranged in front of the outlet ports and a polarisation state analyser (PSA), at least some of the optical elements of the analyser being arranged in front of one of the inlet ports (211b), the analyser comprising at least one optical element (222) held by at least one support, the support being applied only to a portion of the contour of the optical element, thereby providing the element with a free edge, which partially overlaps with the other inlet port (211a).

39. The system of claim 38, wherein the support extends over an angle between 180° and 300° in contact with the optical element, preferably presenting a generally C-shape, said shape opening substantially at 45° in a downwardly sloping direction.

40. The system of claim 38 or 39, comprising at least one further support for holding at least one optical element of the polarization state generator placed in front of the light exit port.

41. The system of claim 38, wherein the other support member is generally C-shaped and opens upward.

42. A system as claimed in any one of claims 36 to 39, each support comprising a series of parts holding together optical elements of the polarization state generator or analyser, wherein at least two supports for holding two consecutive optical elements share an intermediate support part.

43. A system as claimed in claim 40, comprising at least two continuous support parts assembled against each other, each of these parts having on one of its faces a housing for receiving a corresponding optical element and one of the parts being used to retain the optical element of the other part.

44. A system as described in any one of claims 38 to 43, wherein the polarization state generator and the analyzer each include similar optical elements placed in opposite order relative to the direction of light propagation, the generator preferably includes a linear polarization filter P, a quarter-wave liquid crystal polarization modulator QFLC, a half-wave plate QWP and a half-wave liquid crystal polarization modulator HFLC in the direction of light propagation, and the analyzer preferably includes a half-wave liquid crystal polarization modulator HFLC, a half-wave plate QWP and a quarter-wave liquid crystal polarization modulator QFLC in the direction of light propagation, and the polarizer (224) associated with the analyzer is preferably deployed upstream of a camera (5) for collecting intensity images to generate a Mueller matrix after returning light from the entrance port to the associated eyepiece.

45. The system of any one of claims 38 to 44, the housing (233) receiving the electrically controllable optical element comprising at least one passage (233a, 233b, 233c) for an electrical cable.

46. ​​The system of any one of claims 38 to 45, being a colposcopy system, comprising: - a base with wheels (31); - a column (30) supported by a base with wheels; - a head (21) supported by an articulated arm connected to the base; - at least one reference reflector (81) supported by the column (30) for calibrating the polarimetry system, and preferably also comprising: - a workstation including a computer (60) supported by a base with wheels; - a keyboard (41) supported by an articulated arm connected to the column; - A screen (40) supported by an articulated arm connected to the column at a height higher than that of the arm supporting the keyboard.

47. A system as claimed in claim 46, wherein the reference reflector (81) is deployed in a light shield (80) on top of the column, hinged about a vertical axis, comprising a frosted metal wall on one side, preferably made of aluminum, and a reference surface with known spectral characteristics on the opposite side.

48. A system as claimed in any one of claims 46 and 47, comprising a pedal (100) allowing a user to trigger a predefined action, in particular to start polarimetry acquisition, the head (21) comprising a button (101) for triggering another predefined action, in particular for starting and stopping video recording.

49. The system of any one of claims 38 to 48, being a colposcopy system, comprising an illumination system, said illumination system comprising: o at least one light source (50); a liquid light guide (51) connected at one end to a light emitting source, the light guide having a core diameter of 5 mm or less; The optical system has a light inlet port connected to the other end of the liquid light guide.

50. The system of claim 49, the optical system comprising an aspheric lens (180) and a deflecting prism (181) between the light inlet port and the light outlet port.

51. A system as claimed in claim 49 or 50, the liquid light guide (51) having an inner diameter in the range between 2.5 and 3.5 mm.

52. A polarimetric system (90) intended to be attached to the head of an observation system, in particular a colposcopy head, the head comprising an optical system having a light outlet port for illuminating the area to be examined and left and right inlet ports facing the area to be observed, the polarimetric system comprising a polarization state generator placed in front of the outlet port and a polarization state analyzer, at least some of the optical elements of the polarization state analyzer being placed in front of one of the inlet ports, the analyzer comprising at least one optical element supported by at least one support, the support being applied only to a portion of the contour of the optical element (222) thereby providing this optical element with a free edge which is arranged to partially overlap with the other inlet port, the polarimetric system comprising a component (242) for attachment to the head (21) of the observation system.

53. A polarimetric system (90) intended to be attached to a head of an observation system, in particular a colposcopic head, said head comprising an optical system having a light exit port for illuminating an area to be examined and at least one entry port, said polarimetric system comprising a polarization state generator (PSG) placed in front of the exit port and a polarization state analyzer (PSA), at least some of the optical elements of said polarization state analyzer being placed in front of the entry port, wherein at least one of the analyzer and the generator comprises an optical element (222) held by at least one support, Each support comprises a series of parts that hold together the optical elements of the polarization state generator or analyzer, wherein at least two supports are used to hold two consecutive optical elements that share an intermediate support part, each of these parts having a housing on one of its sides for receiving the corresponding optical element and one of the parts is used to retain the optical element of the other part.

Images