Laser microscope with ablation function

Abstract

Laser microscope (1), comprising at least a first laser source (10) emitting at least one, in particular pulsed, excitation beam (11), a scanning optic (3) configured for scanning the excitation beam (11) over the surface (55) of a sample (5), a focusing optic (4) configured for focusing the excitation beam (11) onto the sample (5), and at least one detector (61, 62) for light (7, 7a-7d) emitted by the sample (5) due to an optical effect in response to the excitation beam (11), wherein a second laser source (20) for a pulsed ablation beam (21) is provided for local ablation of the material of the sample (5), wherein the ablation beam (21) is guided to the sample (5) via the scanning optic (3) and the focusing optic (4), characterized in that the first laser source (10) emits pulses (11a, 11b) emits at least two different wavelengths, in particular pulses of three different wavelengths,• the first laser source (10) and the second laser source (20) are fed by a common continuous-wave pump laser (15), • the beam from the common continuous-wave pump laser (15) is guided into an optical oscillator (16) and a beam splitter (18) is provided to divide the pulsed beam (17) emitted by the optical oscillator (16) into the excitation beam (11) on the one hand and the ablation beam (21) on the other hand, and • the excitation beam (11) is guided through a spectral filter (19).

Description

The invention relates to a laser microscope with which a sample can be both examined and selectively modified by removing material, and to a method for operating it. State of the art In laser microscopy, a laser beam is scanned across the surface of the sample under investigation using scanning optics. The spatial resolution of the illuminated area on the sample is diffraction-limited. The use of nonlinear effects for imaging necessitates a transition from continuous to pulsed illumination. These effects depend quadratically or even more strongly on the light intensity, meaning they only produce a usable signal above a certain minimum intensity. Maintaining this intensity continuously would require a very high level of technical effort and would also destroy the sample through heating. Therefore, the laser energy is concentrated in short pulses with high instantaneous intensity, with the average power deposited in the sample carefully selected to prevent excessive heating. Such laser microscopes are described, for example, in (T. Meyer, M. Baumgartl, T. Gottschall, T. Pascher, A. Wuttig, C. Matthäus, BFM Romeike, BR Brehm, J. Limpert, A. Tünnermann, O. Guntinas-Lichius, B. Dietzek, M. Schmitt, J. Popp, “A compact microscope setup for multimodal nonlinear imaging in clinics and its application to disease diagnostics”, Analyst 138 (14), 4048-57 (2013)) and (T. Meyer, M. Chemnitz, M. Baumgartl, T. Gottschall, T. Pascher, C. Matthäus, BFM Romeike, BR Brehm, J. Limpert, A. Tünnermann, M. Schmitt, B. Dietzek, J. Popp, “Expanding Multimodal Microscopy by High Spectral Resolution Coherent Anti-Stokes Raman Scattering Imaging for Clinical Disease Diagnostics”, Analytical Chemistry 85, 6703-6715 (2013 )).These microscopes combine coherent anti-Stokes Raman scattering (CARS), two-photon excited fluorescence (TPEF), and second-harmonic generation (SHG). Such multimodal imaging is advantageous in two respects: First, it allows for the generation of molecule-specific contrast, which can be used, for example, to differentiate pathologically altered tissue from healthy tissue in clinical diagnostics. Second, the non-linear dependence of these effects on light intensity means that, of the typically Gaussian-distributed spatial intensity profile of the pulse, only the center with the highest intensity contributes to the image. The spatial resolution is therefore better than would be expected based on the diffraction limit. US 2010 / 286 674 A1 discloses a system for imaging and simultaneous surgical manipulation of biological tissue using an ultrashort pulsed ablation laser and an imaging light source. US 6,166,385 A discloses a scanning laser microscope based on the simultaneous absorption of multiple photons of the excitation light and subsequent emission of fluorescence light. This effect is highly intensity-dependent and thus essentially limited to the vicinity of the focal plane of the excitation light. DE 10 2014 110 575 A1 discloses a microscope in which an observation beam path for observing a sample and an illumination beam path for illuminating and / or manipulating the sample are combined via a beam splitter. DE 10 2006 039 083 A1 discloses a tunable illumination source for microscopes. Task and solution The object of the present invention is to extend known laser microscopes by the possibility of modifying the sample locally with high precision and thereby realizing a structure that is as compact as possible and therefore suitable for clinical use. This problem is solved according to the invention by a laser microscope according to the main claim and by a method for operating it according to the dependent claim. Further advantageous embodiments are described in the dependent claims relating thereto. Subject matter of the invention Within the scope of the invention, a laser microscope was developed. This laser microscope comprises at least one first laser source emitting at least one, in particular pulsed, excitation beam; scanning optics configured for scanning the excitation beam across the surface of a sample; focusing optics configured for focusing the excitation beam onto the sample; and at least one detector for light emitted by the sample in response to the excitation beam due to an optical effect. The laser microscope can, for example, be configured for multimodal imaging. The optical effect can be linear. Imaging can then occur particularly quickly because a high signal intensity is available. However, a nonlinear optical effect is especially advantageous; that is, the detector is sensitive to light emitted by the sample in response to the excitation beam due to a nonlinear optical effect. This response then originates primarily from the central region of the excitation beam's beam profile, where the instantaneous intensity is at its maximum. According to the invention, a second laser source for a pulsed ablation beam is provided for the purpose of local ablation of the sample material, wherein the ablation beam is guided to the sample via the scanning optics and the focusing optics. It was recognized that this combination makes it possible to select structures from an imaged area and ablate only these structures with high selectivity. The fact that ablation requires a significantly higher intensity than imaging does not necessarily mean that the spatial resolution during ablation is lower than during imaging. By appropriately selecting the laser parameters, the ablation beam pulses can be designed to interact directly with the electron shells of the atoms in the sample material, ionizing them. This locally vaporizes the sample material by converting the electrons into a plasma. If the ablation pulse is sufficiently short, it interacts with the sample material only in this way, so that, in particular, no heat is deposited in the sample in the form of excitation states.The ionization of the electron shells requires such a high instantaneous intensity that it exists only at the immediate center of the spatial intensity distribution of the ablation pulse. Therefore, the spatial resolution of the ablation is at least as good as, if not better than, the spatial resolution of the imaging. Both the excitation beam and the ablation beam exhibit an inhomogeneous intensity distribution not only laterally, i.e., in the plane perpendicular to the direction of propagation, but also along the direction of propagation. This results in a highly localized center of maximum intensity, both laterally and along the direction of propagation. Consequently, the imaged or ablated area can be selected not only laterally but also with respect to its depth below the sample surface. This allows, for example, the investigation and targeted modification of structures within a biological sample without the need for destructively opening the sample surface at the location of these structures. Both the excitation and ablation beams can penetrate the sample to depths of up to several hundred micrometers. The shared guidance of the excitation and ablation beams through the same scanning and focusing optics minimizes any systematic offset between the points where both beams arrive at the sample. It also minimizes the effort required for adjustment. As a result, successful use of the laser microscope no longer requires the user to be an expert in laser microscopy. Instead, the laser microscope, including the new ablation function, is also accessible to users who are only experts in image interpretation, such as physicians or biologists in clinical diagnostics. For such applications in particular, it is advantageous that the shared use of the scanning and focusing optics for both excitation and ablation allows for the integration of both functions into a single compact device. In nonlinear imaging, photons from the excitation beam are used to switch excitation states in the sample material. For this to occur, the energy of the photon must match the energy difference between the excitation states. Therefore, this imaging technique requires an excitation beam with one or more specific wavelengths, which are tailored to the sample material and the effect to be used for imaging. The inventors recognized that, in contrast, nonlinear ablation is essentially independent of the wavelength of the ablation beam. The partial ionization of the electron shells of atoms in the sample material is caused directly by the instantaneous electric field acting on the electrons. The oscillation frequency of this electric field, and thus the wavelength of the ablation beam, is therefore irrelevant. This wavelength can thus be freely selected based on practical or instrumental considerations. The fact that the excitation beam and the ablation beam interact with the sample in qualitatively completely different ways necessitates that the pulses of the ablation beam must be significantly shorter than the pulses of the excitation beam. The maximum instantaneous intensity of an ablation pulse is typically about 1000 times greater than the maximum instantaneous intensity of an excitation pulse. Accordingly, an ablation pulse can, for example, have a pulse energy in the range of 0.1 µJ to 10 µJ with a pulse duration of 100 fs, while an excitation pulse, for example, can only have a pulse energy in the range of 1 nJ to 10 nJ with a pulse duration of approximately 10 ps. A particular advantage of integrating the laser microscope and ablation tool into a single device is that the ablation process can be interrupted at any time and visually monitored by taking a new microscope image. This allows for real-time control of the selectivity with which the sample material is removed. In a particularly advantageous embodiment of the invention, at least one wavelength emitted by the second laser source coincides with at least one wavelength emitted by the first laser source. The refraction of light at the scanning optics, as well as the focusing of light by the focusing optics, is wavelength-dependent. Therefore, an excitation beam and an ablation beam with different wavelengths, guided into the scanning optics in a common beam path, can be chromatically shifted relative to each other and arrive at the sample with a spatial offset. This chromatic shift is minimized when the wavelengths of both beams are identical. Alternatively, the two beams can have different wavelengths. They can then be combined, particularly via a dichroic beam splitter, with only minimal intensity loss. In a further particularly advantageous embodiment of the invention, the polarization directions of the first and second laser sources enclose an angle between 70 and 110 degrees. Preferably, both polarization directions are orthogonal to each other. The excitation beam and the ablation beam can then be combined, particularly via a polarization-preserving beam splitter, with only minimal intensity losses. The interaction of both beams, especially with biological samples that do not exhibit a crystalline preferred orientation, is generally independent of the polarization direction. Furthermore, different polarization directions of the excitation beam and the ablation beam do not lead to an offset between the points where both beams arrive on the sample when passing through the scanning and focusing optics. The first and second laser sources are fed by a common continuous-wave pump laser. The beam from the common pump laser is guided into an optical oscillator, and a beam splitter is provided to divide the pulsed beam emitted by the optical oscillator into the excitation beam on the one hand and the ablation beam on the other. By sharing components for the excitation beam and the ablation beam, costs, installation space, and energy consumption can be reduced. Furthermore, adjustment is significantly simplified. If, due to market availability, two separate optical oscillators for the two energetically very different beams are more cost-effective than a single oscillator suitable for both beams, then it may also be advantageous to use two separate oscillators. Furthermore, the excitation beam is guided through a spectral filter. If, for example, the pulsed beam emitted by the optical oscillator has the very short pulse duration intended for the ablation beam, the spectral filter, due to Heisenberg's uncertainty principle, causes the pulses of the excitation beam to be significantly lengthened. At the same time, the spectral filter can also block components of the excitation beam that are not suitable for changing excitation states in the sample and thus only contribute to heating the sample. The first laser source emits pulses of at least two different wavelengths. In particular, the first laser source can emit pulses of three different wavelengths. Such a laser source is especially suitable for coherent anti-Stokes Raman scattering. For this purpose, two of the emitted wavelengths advantageously have a difference that is suitable for exciting at least one vibrational state in a molecule of the sample material. For example, a first emitted wavelength can be tunable in the range between 1025 nm and 1075 nm, and a second emitted wavelength can be tunable in the range between 800 nm and 1000 nm. The different wavelengths can be generated, for example, by a four-wave mixture of wavelengths symmetrically distributed around the wavelength of a pump laser used as an energy source. Ytterbium-doped fiber lasers are suitable for this purpose.The second laser source can, for example, also be such a fiber laser, whose wavelength is tunable in the range between 1030 nm and 1060 nm. The detector is thus advantageously designed to detect light formed from the excitation beam by coherent Raman scattering, in particular by anti-Stokes Raman scattering. In a further particularly advantageous embodiment of the invention, the wavelength emitted by the first laser source and / or the second laser source is between 750 nm and 3 µm, preferably between 750 nm and 1.5 µm. This wavelength range is particularly advantageous for the examination and modification of biological samples, since scattering losses in the tissue are minimized and the light can penetrate to a depth of several hundred µm into the sample. The invention also relates to a method for operating a laser microscope, wherein in the laser microscope an excitation beam, in particular a pulsed beam, and a pulsed ablation beam are directed towards a sample, wherein means are provided for scanning the excitation beam and the ablation beam across the sample. The laser microscope further includes at least one detector for light emitted by the sample due to a nonlinear optical effect in response to the excitation beam. According to the invention, the pulse duration of the ablation beam is selected between 35 fs and 300 fs, preferably between 100 fs and 300 fs. A first laser source for the excitation beam emits pulses of at least two different wavelengths, in particular pulses of three different wavelengths. The first laser source and a second laser source for the ablation beam are fed by a common continuous-wave pump laser. The beam from the common continuous-wave pump laser is fed into an optical oscillator. A beam splitter is provided to divide the pulsed beam emitted by the optical oscillator into the excitation beam on the one hand and the ablation beam on the other. The excitation beam is passed through a spectral filter. It was discovered that, particularly with a pulse duration in this range, local ablation of the sample material can occur without the sample being heated excessively. As previously explained, the sample material is vaporized by the instantaneous electric field of the ablation pulse, which partially ionizes the electron shells of atoms in the sample material. This effect only occurs above a certain minimum field strength, sufficient to overcome the binding energy of at least the outer electrons. This minimum field strength corresponds to a minimum value for the instantaneous intensity of the ablation pulse (on the order of 10¹²–10¹⁴ W / cm²).The instantaneous intensity must reach this minimum value so rapidly during the ablation pulse, rising on a rising edge, and then falling again on a falling edge at the end of the pulse, that no other interaction, particularly thermal, occurs between the ablation pulse and the sample material on these edges. The ablation pulse must therefore rise and fall on a faster timescale than is necessary to excite vibrations or rotations in the molecules of the sample material and thus couple heat into it. If such excitation of vibrations occurs, the sample will very likely be heated to such an extent that it is destroyed.Selective ablation is based on the fact that during the pulse phase in which direct ionization of the electron shells of atoms takes place, at least one order of magnitude more energy is coupled into the sample than during the rising and falling edges of the pulse, during which the instantaneous intensity is insufficient for direct ionization. When sample material is locally ablated using pulses according to the invention, this can be achieved, for example, with average ablation beam powers on the order of 1 mW. In a particularly advantageous embodiment of the invention, the pulse duration of the excitation beam is selected to be 10 to 1000 times longer than the pulse duration of the ablation beam. This ensures that, firstly, the excitation beam does not ablate material through direct ionization, and secondly, that sufficient time is available to generate a specific excitation state in the sample through the interaction of the photons of the excitation beam with the sample. The qualitative difference between the effects of the excitation beam and the ablation beam is primarily due to the different timescales and intensity scales on which these effects occur. Advantageously, the pulse duration of the excitation beam is selected from a range between 1 ps and 100 ps, ​​preferably between 10 ps and 40 ps, ​​and most preferably between 20 ps and 40 ps. Adjusting the excitation beam is simplest in this range. Furthermore, the ranges between 10 ps and 40 ps, ​​and between 20 ps and 40 ps, ​​are particularly advantageous when the excitation beam is guided through at least one optical fiber, for example, when the first laser source is a fiber laser. Optical fibers typically have a dispersion of approximately 10 ps per meter of length. From a pulse duration of approximately 10 ps, ​​the dispersion in the optical fiber, combined with the dispersions in the scanning optics and the focusing optics of the microscope, is low enough to no longer significantly influence the choreography of a spectroscopy with excitation by a pump pulse and interrogation by an interrogation pulse (pump-probe spectroscopy), when parts of the microscope optics, such as the...Objectives, scan lenses, tube lenses, or condensers can be changed. Furthermore, the range between 20 ps and 40 ps is optimal with regard to the spectral resolution, which is directly linked to the pulse duration by Heisenberg's uncertainty principle. Advantageously, the pulse repetition rate of the excitation beam is chosen to be between 1 MHz and 40 MHz, preferably between 1 MHz and 20 MHz. This range represents an optimal compromise between the highest possible image acquisition speed on the one hand and the lowest possible sample heating on the other. While in ablation the energy coupled into the sample is essentially carried away directly with the vaporized material, leaving hardly any heat in the sample, the excitation beam heats the sample according to its average power. At least one pulse of the excitation beam is required to acquire each image pixel. Depending on the signal-to-noise ratio of the effect chosen for imaging, it can also be advantageous to provide several pulses of the excitation beam per image pixel to obtain better statistics. Advantageously, the repetition rate of the ablation beam pulses is chosen to be between 100 kHz and 10 MHz, preferably between 100 kHz and 1 MHz. For ablating larger structures quickly, the ablation beam can be configured, for example, so that each pulse is effective in an area encompassing several image pixels, such as approximately 10 pixels, of the image acquired with the excitation beam. The ablation beam can then be scanned more quickly, i.e., in a wider grid of grid points. Ideally, the areas in which each ablation pulse removes material from the sample seamlessly combine to form the structure to be ablated. During the ablation process, the focal area of ​​the ablation beam can also be varied in size, for example, to first ablate large-area structures at high speed and then rework fine structures with greater accuracy. The image field of the laser microscope can, for example, have an area of ​​1 mm². A spatial resolution of better than 1 µm laterally, i.e., along the surface of the sample, is typically achievable. Axially, i.e., at depths below the surface of the sample, a resolution of better than 5 µm is typically achievable. The ablation beam can remove material with a resolution of typically around 1 µm³. In a further particularly advantageous embodiment of the invention, the image obtained by scanning the excitation beam is evaluated by applying at least one multivariate classifier to determine whether the sample exhibits a predetermined structure or property. For this purpose, the classifiers disclosed in European patent application 15 200 864.5, which originates from the applicants, can be used, for example. For many applications, a large selection of multivariate classifiers is available. From this selection, the classifiers to be used can be chosen, for example, according to the required evaluation time, in order to complete the evaluation within a given timeframe. Furthermore, in biological in vivo applications where motion artifacts can occur, there may be requirements for a minimum raster speed to minimize these artifacts. Increased speeds can then lead to increased noise. This noise can affect different multivariate classifiers to varying degrees. To assess the reliability of potential classifiers under the influence of image noise, the image obtained by rasterizing the excitation beam is modified in a further particularly advantageous embodiment of the invention by superimposing noise to create a test image. The reliability of the classifier is then evaluated by comparing the results obtained when applied to the image on the one hand and to the test image on the other. For example, a classifier that changes its opinion even with minimal additional noise can be considered less reliable than a classifier that only changes its opinion with very high levels of additional noise. The reliability determined in this way will typically depend on the type and intensity of the noise contained in the image acquired by the laser microscope. The noise, in turn, depends on the speed of image acquisition. By making the reliability quantitatively verifiable through the addition of extra noise, the laser microscope user can choose an optimal compromise between the speed of image acquisition on the one hand and the availability of as many informative classifiers as possible on the other. The method is particularly advantageous when carried out with a laser microscope of the invention. The laser microscope is specifically designed to be operated with the method according to the invention, and vice versa. One possible application of the laser microscope and the method according to the invention is CARS-guided femtosecond laser ablation of tissue for microsurgical procedures. This application is based on the combination of multimodal nonlinear microscopy (CARS, TPEF, SHG) or coherent anti-Stokes Raman scattering microscopy (CARS microscopy) of tissue for local diagnostics and characterization with the targeted ablation of tissue portions by femtosecond laser ablation. Key components of this application are the data acquisition and processing method, the combination of the imaging technique with a tissue ablation method, and the laser source used for this purpose. The method enables imaging and molecule-sensitive detection of target structures in the tissue without the use of external marker substances and the subsequent precise ablation of these target structures.It allows for non-destructive, three-dimensional imaging of tissue down to several hundred micrometers in depth, and the targeted removal of tissue down to several hundred micrometers below the surface, i.e., without creating an open wound. This significantly reduces the risk of infection. The procedure is suitable for all body regions accessible to microscopes, such as the skin, as well as for surgical procedures performed with operating microscopes, e.g., in the ear, nose, and throat area, or with flexible or rigid imaging endoscopes. This application combines marker-free molecular imaging for localizing disease-related tissue abnormalities with femtosecond laser ablation for targeted removal. This combines diagnostics and therapy in a single device, contributing to faster treatment. Furthermore, femtosecond laser ablation allows for significantly more precise removal of target structures and, in the future, can also be used in endoscopes and microendoscopes. This makes the procedure particularly advantageous in the vicinity of physiologically important tissue structures, such as in the larynx near the vocal cords or in the brain. The previous gold standard relied on taking and histologically processing tissue biopsies to diagnose the disease and potentially perform further surgery if the examination confirmed the suspicion of a serious condition. Thin tissue sections were prepared from the biopsy material and stained histologically, primarily using hematoxylin and eosin staining. The stained tissue section was then evaluated by a pathologist. This established process was time-consuming and could take several days. The accuracy of conventional surgical procedures was limited to approximately 100 µm. Due to the time-consuming sample processing, the success of the surgery could not be verified during the operation, sometimes necessitating costly repeat surgeries. In most cases, a significant amount of tissue was removed, which increased the risk of infection and could damage important physiological structures. The invention combines a marker-free imaging technique, capable of directly visualizing molecules, with an optical method for tissue ablation. Specific laser parameters enable a high penetration depth of several hundred micrometers for both imaging and laser ablation. This method is significantly faster and more precise than conventional techniques. The new workflow using the invention comprises: (i) visualization of the target region using multimodal nonlinear microscopy, e.g., coherent anti-Stokes Raman scattering microscopy (at one or more vibrational frequencies) alone or in combination with two-photon fluorescence and the second harmonic; (ii) multivariate analysis of the image data to identify the target region for laser ablation (based, for example, on the previous European patent application 15 200 864.5); and (iii) local ablation of target tissue and tissue structures in vivo / excorpore-in vivo / in vitro / ex vivo, even below an intact tissue layer. The key innovation in this workflow is the combination of tissue measurement and imaging using multimodal nonlinear microscopy (e.g., CARS, TPEF, SHG) with laser ablation for targeted tissue removal. The compact laser source for CARS imaging and laser ablation is the most important tool in this context. Both the combination of these two methods and the design of a compact, air-cooled, high-performance ablation laser are fundamentally new. Furthermore, neither coherent Raman microscopy nor multimodal nonlinear microscopy has yet been used intraoperatively. Applications in animal experiments are also limited to imaging. Coupling these techniques with optical methods for targeted tissue ablation and online monitoring of surgical progress is entirely new. In comparison to the use of coherent Raman microscopy and multimodal nonlinear microscopy for frozen section diagnostics, for example according to European patent application 15 200 864, which is attributed to the inventors.5. The following differences are important: - Examination of extensive intact tissue samples, not thin frozen sections on slides - Detection of signals in reflection: since signal detection in the forward direction is not possible for extensive tissue structures, the signals must be recorded in the reverse direction - Real-time examination of tissue: since motion artifacts occur, the examinations are performed at a higher speed compared to histological frozen section diagnostics, therefore higher noise is to be expected and the automated data analysis is limited to a few important parameters - Real-time analysis: online data processing directly after data acquisition - Excitation: NIR laser, 750-1500 nm - to achieve a high penetration depth, long-wavelength illumination is chosen for imaging and femtosecond ablation, in order to minimize scattering losses in the tissue. The new workflow, in which the laser microscope and the method according to the invention are the essential tools, has the following key advantages: • Pathological tissue structures can be detected and visualized in vivo, thus making their demarcation from the surrounding healthy tissue visible. • Pathological tissue structures can be selectively removed with micrometer spatial resolution, even in 3D and surrounded by healthy tissue. • The new workflow is also suitable for critical operations on physiologically important structures, as it allows for extremely high precision and enables high-contrast visualization of target structures. Contrast agents are not required. • Since the examination can be performed directly in the operating room, the method saves time and money, as biopsy sampling and evaluation are unnecessary.Since the success of the surgery can be checked immediately, repeat surgeries can be avoided, which can lead to significant cost savings in surgical patient care. Special description section The subject matter of the invention is explained below with reference to the figures, without thereby limiting the subject matter of the invention. Figure 1 shows an embodiment of the laser microscope 1 with two separate laser sources 10 and 20 for excitation and ablation, a CARS detector 61 in transmission, and a CARS detector 62 in reflection. Figure 2 shows selective ablation of a deposit 82 from an arterial wall 81. Figure 3 shows a further embodiment of the laser microscope 1 with a common laser source 10 / 20 for excitation and ablation and a multi-stage CARS detector 62 in reflection. Figure 4 shows testing the reliability 31a-39a of a multivariate classifier 31-39 by superimposition with test noise 65. Fig. 1 shows a first embodiment of a laser microscope 1 according to the invention. A first laser source 10 emits an excitation beam 11 consisting of pulses 11a of a first wavelength and pulses 11b of a second wavelength, the difference between the wavelengths of pulses 11a and 11b corresponding to the frequency of an oscillation in biological sample material 5a. A dichroic beam splitter 91 directs the excitation beam 11 towards the scanning optics 3. A second laser source 20 emits an ablation beam 21 consisting of pulses of a further wavelength. The ablation beam 21 is also guided into the scanning optics 3 via a mirror 22 and the dichroic beam splitter 91. From the scanning optics 3, the excitation beam 11 and the ablation beam 21 are guided together into the focusing optics 4, which comprises a scanning and tube lens system 4a, a further dichroic beam splitter 4b, and an objective 4c. The beams 11 and 21 are focused together onto the biological sample material 5a, which is applied as a thin layer to a microscope slide 5b and, together with the slide 5b, forms the sample 5. The surface 55 of the sample 5 is approximately planar. A portion of the light 11a, 11b, 21 incident on sample 5, as well as the coherent Raman-scattered light 7 from sample 5, is transmitted and enters a first multimodal CARS detector 61. In the CARS detector 61, all the light 7, 11a, 11b, 21 first passes through a condenser 61a and is converted into a parallel beam path. A dichroic beam splitter 61b separates the coherent anti-Stokes Raman scattered light 7a and directs it, via a dielectric filter 63a, which retains in particular residual components of all laser beams 11a, 11b, and 21, to a first photomultiplier 61c. The light transmitted by the beam splitter 61b contains a further signal component 7b, which is due to two-photon excitation fluorescence (TPEF), second harmonic generation (SHG) or another freely selectable optical effect.This signal component 7b is separated from the laser beams 11a, 11b, 21 by a further dielectric filter 63b and directed to a second photomultiplier 61d. If the dielectric filter 63b is removed, the laser light 11a, 11b, 21 can optionally be monitored for intensity fluctuations using the second photomultiplier 61d. The dielectric filters 63a and 63b typically have an optical density of around 6 for the laser wavelengths used. They can optionally be supplemented by a further short-pass filter, not shown in Fig. 1, which is arranged between the condenser 61a and the dichroic beam splitter 61b. The light 7, 11a, 11b, 21 reflected from the sample passes through the beam splitter 4b and enters the second multimodal CARS detector 62. In the second CARS detector 62, the Raman-scattered component 7a of the light is separated by a dichroic beam splitter 62x and directed via a dielectric filter 63c, which specifically retains residual components of all laser beams 11a, 11b, and 21, to a photomultiplier 62y. Analogous to the first CARS detector 61, which operates in transmission configuration, the light transmitted by the beam splitter 62x contains the signal component 7b. This signal component 7b is separated from the laser beams 11a, 11b, 21 by another dielectric filter 63d and is directed to a photodiode 62z. If the dielectric filter 63d is removed, the photodiode 62z can be used to monitor the laser beams 11a, 11b and 21 for intensity fluctuations or, for example, to normalize the Raman spectra to the overall intensity.The 62z photodiode is better suited for this purpose than a 61a, 61b, 62y photomultiplier due to its larger dynamic range. Conventional photomultipliers with secondary electron multipliers can be used as photomultipliers 61a, 61b, and 62y. Alternatively, hybrid detectors can be used. In such hybrid detectors, primary electrons are generated in a cathode, which can consist, for example, of gallium arsenide phosphide. The primary electrons are then accelerated by a significantly higher voltage (approximately 5–10 kV) compared to conventional photomultipliers onto a material that releases secondary electrons. The secondary electrons are then directed onto a diode and converted by this diode into a current pulse. Fig. 2 illustrates the selective ablation with the ablation beam 21. Fig. 2a schematically shows a first image 64 of a thin section of an arterial wall 81, taken with the laser microscope 1 according to Fig. 1. Deposits 82 have accumulated on the inner surface of the arterial wall 81 inside 83 of the artery. Fig. 2b schematically shows another image 64 of the same image field after the selective removal of the deposits 82 with the ablation beam 21. The arterial wall 81 itself is undamaged. The real CARS recordings underlying the schematic Fig. 2, for a vibrational resonance of 2850 cm⁻¹, are to be understood as a "proof of concept" for the fundamental feasibility of selective tissue ablation. In actual in vivo applications, the sample, including the structures to be ablated, is not presented as a thin section, but as a three-dimensional object. A laser microscope 1 adapted to real in vivo applications is sketched in Fig. 3a. In contrast to Fig. 1, here a single laser 10=20 is the common source for the excitation beam 11 and the ablation beam 21. This common laser 10=20 is significantly more compact than the arrangement of two separate lasers 10, 20 according to Fig. 1. Furthermore, in contrast to Fig. 1, the ablation beam 21 has a wavelength that is also contained in the excitation beam 11. However, the polarization direction of the ablation beam 21 is orthogonal to the polarization direction of the excitation beam 11. Therefore, the excitation beam 11 and the ablation beam 21 are combined via a polarization-preserving beam splitter 92. The excitation beam 11 and the ablation beam 21 are guided to the sample 5 via the common scanning optics 3 and the common focusing optics 4, analogous to Fig. 1. The sample 5 is a three-dimensional tissue object 5d on which a structure 5c to be ablated is indicated. Accordingly, the surface 55 of the sample 5 is also three-dimensional. Unlike in Fig. 1, sample 5 is not transparent. Therefore, measurements can only be taken in reflection. The light 11a, 11b, 21 reflected from sample 5, together with the signal light 7 generated by sample 5, passes through the beam splitter 4b of the focusing optics 4 and enters the single multimodal CARS detector 62. In this CARS detector 62, the various signals, i.e., the Raman-scattered light 7a, SHG signals 7b, TPEF signals 7c, another signal component 7d, and the laser light 11a, 11b, 21, are separated by several cascaded dichroic beam splitters 62a, 62b, and 62d, as well as matching dielectric filters 63a, 63b, 63c, and 63d. The first dichroic beam splitter 62a splits off a first wavelength component 7a of the signal light 7 and directs it via the dielectric filter 63a to the photomultiplier 62f. The remaining wavelength components 7b and 7c, e.g. TPEF and SHG, the reflected excitation light 11a, 11b and the reflected ablation beam 21 pass through the first dichroic beam splitter 62a unhindered in the forward direction (vertically upwards in Fig. 3a). The second dichroic beam splitter 62b splits off a second wavelength component 7b and a third wavelength component 7c of the signal light 7. These two wavelength components 7b and 7c are then separated from each other in a third dichroic beam splitter 62d and directed to the photomultipliers 62e and 62c, respectively, via dielectric filters 63b and 63c, which each allow only wavelength component 7b or 7c to pass through and block other spectral components. The reflected excitation light 11a, 11b and the reflected ablation beam 21 pass through the second dichroic beam splitter 62b unimpeded, along with another signal component 7d. The dielectric filter 63d blocks the laser light 11a, 11b, 21, so that only the signal component 7d reaches the photodiode 62q. Optionally, the dielectric filter 63d can be removed so that the photodiode 62q can be used to measure the intensity of the laser light 11a, 11b, 21.This intensity can then be used, analogous to Fig. 1, to check and normalize the Raman and other nonlinear signals to the overall intensity. The advantage of the CARS detector 62 according to Fig. 3 is that it can simultaneously register four wavelength components 7a, 7b, 7c and 7d of the signal light 7, e.g. CARS, SHG, TPEF and another freely selectable signal component. These four wavelength components 7a, 7b, 7c and 7d can be generated by the sample truly simultaneously. However, they can also be generated sequentially, for example by tuning the wavelengths of the pulses 11a and 11b that form the excitation beam 11. Fig. 3b shows the internal structure of the common laser 10=20. This laser 10=20 is characterized by the fact that most of the optical components are used for both the excitation beam 11 and the ablation beam 21. The beam from a common continuous-wave pump laser 15 is guided into an optical oscillator 16 and is converted there into pulses with a pulse duration suitable for the ablation beam 21. The beam 17 formed from these pulses is guided from the optical oscillator 16 to a beam splitter 18. The beam splitter 18 allows the ablation beam 21 to pass through in a forward direction (vertically upwards in Fig. 3b). The ablation beam 21 is amplified by an amplifier 18b and finally exits the laser 10=20. The excitation beam 11 is directed laterally to a mirror 18a and from there to a spectral filter 19. Due to Heisenberg's uncertainty principle, the spectral filter 19 significantly lengthens the pulses of the excitation beam 11. The excitation beam 11 is then amplified in an amplifier 19a. In a photonic crystal fiber 19b, two further wavelengths, signal and idler, are generated from the excitation beam 11, which essentially has only one frequency ω0 behind the spectral filter 19, by four-wave mixing. Through the nonlinear process of four-wave mixing, two photons of the excitation beam with frequency ω0 generate a pair consisting of a signal photon with frequency ω0+Δω and an idler photon with frequency ω0-Δω. The photonic crystal fiber 19b is microstructured in such a way that energy and momentum conservation is maintained despite the dispersion within the crystal fiber 19b. In the photonic crystal fiber 19b, broadband photon pairs (ω0±Δω) with many frequency shifts Δω are generated. To ensure that exactly one frequency shift Δω is favored, and thus that 10 = 20 pulses 11a, 11b with exactly two frequencies (and therefore two wavelengths) emerge from the laser, a portion of the light exiting the photonic crystal fiber 19b is fed back into the photonic crystal fiber 19b via a resonant cavity 19c. The cavity 19c is always resonant at only one frequency, i.e., either at the frequency ω0+Δω or at the frequency ω0-Δω. By thus defining the frequency shift Δω, both frequencies of the photon pair ω0±Δω that are to be preferentially generated are determined. The frequency shift Δω can be determined via the resonance of the cavity 19c. to be voted on. Fig. 4 schematically shows how classifiers 31-39 can be evaluated to determine their suitability for detecting a given structure or property 41-49 in a specific image 64 containing noise 64a, which was acquired with the laser microscope 1. The image 64 is modified to a test image 66 by adding test noise 65. The classifier 31-39 is then applied to the original image 64 and yields a result 67 indicating whether the structure or property 41-49 is present on or in the sample 5, as shown in the original image 64. Simultaneously, the classifier 31-39 is applied to the test image 66 and yields a result 68. The two results are compared in section 69, and the reliability 31a-39a of the classifier 31-39 is evaluated based on this comparison.This reliability 31a-39a can depend in particular on the level of additional noise 65a at which the classifier 31-39 changes its opinion. If even a slight amount of additional noise is sufficient for this, it can be concluded that the original noise 64a in image 64 may already have distorted the result 67 provided by the classifier 31-39. If, on the other hand, the classifier 31-39 does not change its opinion even with strong noise, it can be concluded that the classifier is particularly resistant to noise and therefore particularly reliable. Reference symbol list 1 Laser microscope 10 First laser source for excitation beam 11 11 Excitation beam 11a, 11b Pulses of beam 11 with different wavelengths 15 Common continuous-wave pump laser for lasers 10, 20 16 Optical oscillator 17 Beam from optical oscillator 16 18 Beam splitter for splitting beam 17 into beams 11, 21 18a Mirror for deflecting the excitation beam 11 18b Amplifier for ablation beam 21 19 Spectral filter for excitation beam 11 19a Amplifier for excitation beam 11 19b Photonic crystal fiber for forming photon pairs ω0±Δω 19c Resonant cavity for selecting a frequency shift Δω 20 Second laser source for ablation beam 21 21 Ablation beam 22 Mirror for ablation beam 21 3 Scan optics or means for rasterizing 31-39 Multivariate classifiers 31a-39a Reliabilities of the multivariate classifiers 31-39 4 Focusing optics 4a Scan and tube lens system 4b Beam splitter in focusing optics 4 4c Objective 41-49 Properties,to which the classifiers 31-39 are sensitive 5 Sample 5a Thin section of biological material 5b Microscope slide 5c Structure to be ablated on sample 5 5d Three-dimensional object as sample 5 55 Surface of the sample 5 61 CARS detector in transmission 61a Condenser 61b Beam splitter in CARS detector 61 61c, 61d Photomultiplier in CARS detector 61 62 CARS detector in reflection 62a First dichroic beam splitter for splitting off 7a 62b Second dichroic beam splitter for splitting off 7b, 7c 62c Photomultiplier for wavelength component 7c 62d Third dichroic beam splitter,separates 7b from 7c 62e Photomultiplier for wavelength component 7b 62f Photomultiplier for wavelength component 7a 62q Photodiode for wavelength component 7d 62x Beam splitter in simple CARS detector 62 62y Photomultiplier in simple CARS detector 62 62z Photodiode in simple CARS detector 62 63a-63c Dielectric filters 64 Image taken by laser microscope 1 64a Noise in image 64 65 Test noise 66 Test image generated from image 64 and test noise 65 67 Result of classifier 31-39 on image 64 68 Result of classifier 31-39 on test image 66 69 Comparison of results 67, 68 7 Response generated by sample 5 7a-7d Wavelength components of the response 7 81 Arterial wall 82 Deposits on arterial wall 81 83 Interior of the artery, bounded by arterial wall 81 91 Dichromatic beam splitter for combining 11, 21 92 Polarization-preserving beam splitter for combining 11,21 ω0 Frequency of the excitation beam 11 behind spectral filter 19 Δω Frequency shift in photonic crystal fiber 19b,

Claims

Laser microscope (1), comprising at least one first laser source (10) emitting at least one, in particular pulsed, excitation beam (11), a scanning optic (3) configured for scanning the excitation beam (11) over the surface (55) of a sample (5), a focusing optic (4) configured for focusing the excitation beam (11) onto the sample (5), and at least one detector (61, 62) for light (7, 7a-7d) emitted by the sample (5) due to an optical effect in response to the excitation beam (11), wherein a second laser source (20) for a pulsed ablation beam (21) is provided for local ablation of the material of the sample (5), the ablation beam (21) being guided to the sample (5) via the scanning optic (3) and the focusing optic (4), characterized in that the first laser source (10) emits pulses (11a, 11b) emits at least two different wavelengths, in particular pulses of three different wavelengths,• the first laser source (10) and the second laser source (20) are fed by a common continuous-wave pump laser (15), • the beam from the common continuous-wave pump laser (15) is guided into an optical oscillator (16) and a beam splitter (18) is provided to divide the pulsed beam (17) emitted by the optical oscillator (16) into the excitation beam (11) on the one hand and the ablation beam (21) on the other hand, and • the excitation beam (11) is guided through a spectral filter (19). Laser microscope (1) according to claim 1, characterized in that at least one wavelength emitted by the second laser source (20) is identical to at least one wavelength emitted by the first laser source (10). Laser microscope (1) according to one of claims 1 to 2, characterized in that the polarization directions of the first laser source (10) and the second laser source (20) enclose an angle between 70 and 110 degrees and are preferably orthogonal to each other. Laser microscope (1) according to claim 3, characterized in that the excitation beam (11) and the ablation beam (21) are combined via a polarization-preserving beam splitter (92). Laser microscope according to one of claims 1 to 4, characterized in that the detector (61, 62) is configured to detect light formed from the excitation beam (11) by coherent Raman scattering. Laser microscope (1) according to one of claims 1 to 5, characterized in that the wavelength emitted by the first laser source (10), and / or by the second laser source (20), is between 750 nm and 3 µm, preferably between 750 nm and 1.5 µm. Laser microscope (1) according to one of claims 1 to 6, characterized in that the detector (61, 62) is sensitive to light (7, 7a-7d) emitted by the sample (5) due to a nonlinear optical effect in response to the excitation beam (11). Method for operating a laser microscope (1), wherein in the laser microscope (1) an excitation beam (11), in particular a pulsed beam, and a pulsed ablation beam (21) are directed to a sample (5), wherein means (3) are provided for scanning the excitation beam (11) and the ablation beam (21) over the sample, and wherein at least one detector (61, 62) is provided for light emitted by the sample (5) due to a nonlinear optical effect in response to the excitation beam (11), wherein the pulse duration of the ablation beam (21) is selected between 35 fs and 300 fs, preferably between 100 fs and 300 fs, characterized in that a first laser source (10) emits pulses (11a, 11b) of at least two different wavelengths, in particular pulses of three different wavelengths, for the excitation beam (11).• the first laser source (10) and a second laser source (20) for the ablation beam (21) are fed by a common continuous-wave pump laser (15), • the beam from the common continuous-wave pump laser (15) is guided into an optical oscillator (16) and a beam splitter (18) is provided to divide the pulsed beam (17) emitted by the optical oscillator (16) into the excitation beam (11) on the one hand and the ablation beam (21) on the other, and • the excitation beam (11) is guided through a spectral filter (19). Method according to claim 8, characterized in that the pulse duration of the excitation beam (11) is selected to be longer by a factor between 10 and 1000 than the pulse duration of the ablation beam (21). Method according to one of claims 8 to 9, characterized in that the pulse duration of the excitation beam (11) is selected from a range between 1 ps and 100 ps, ​​preferably between 20 ps and 40 ps. Method according to one of claims 8 to 10, characterized in that the repetition rate of the pulses of the excitation beam (11) is selected between 1 MHz and 40 MHz, preferably between 1 MHz and 20 MHz. Method according to one of claims 8 to 11, characterized in that the repetition rate of the pulses of the ablation beam (21) is selected between 100 kHz and 10 MHz, preferably between 100 kHz and 1 MHz. Method according to one of claims 8 to 12, characterized in that the image (64) obtained by scanning the excitation beam (11) is evaluated by applying at least one multivariate classifier (31-39) to determine whether the sample (5) has a predetermined structure or property (41-49). Method according to claim 13, characterized in that the image (64) is modified to a test image (66) by superimposing noise (65) and the reliability (31a-39a) of the classifier (31-39) is evaluated from the comparison (69) of the results that the classifier (31-39) provides when applied to the image (64) on the one hand (67) and to the test image (66) on the other hand (68).

Images