{
  "id": "optical-microscope",
  "title": "Optical Microscope",
  "lang": "en",
  "source_lang": "en",
  "summary": "An optical microscope uses light and optical components to reveal small structures, supporting visual observation, quantitative measurement, and molecular imaging.",
  "category": "science",
  "subcategory": "physics",
  "type": "concept",
  "aliases": [
    "Light microscope"
  ],
  "generated_at": "2026-11-03",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/optical-microscope/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/optical-microscope/",
      "json": "https://aiwiki.page/zh/optical-microscope.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "An optical microscope is an instrument that uses light to produce an enlarged image or spatially resolved measurement of an object too small to examine adequately with the unaided eye. Most instruments use visible light and glass lenses, although optical microscopy also includes ultraviolet and near-infrared methods, scanning systems, and computationally reconstructed images. Unlike instruments that form images with beams of electrons, optical microscopes can readily examine living specimens in water, follow processes over time, and distinguish selected molecular components through fluorescent labels.\n\nThe optical microscope is a major class of microscope and a foundational instrument in biology, medicine, materials science, and industrial inspection. Its capabilities depend not simply on magnification, but on resolution, contrast, illumination, specimen preparation, and detection. Conventional instruments distinguish details separated by roughly 200–300 nanometres under favorable conditions; specialized fluorescence methods can resolve substantially smaller structures. The development of optical microscopy connects lens craftsmanship and wave optics with cell theory, microbiology, digital imaging, and modern molecular measurement."
    },
    {
      "heading": "Historical development",
      "text": "Early microscopes emerged from the lens-making traditions of late sixteenth- and early seventeenth-century Europe. The invention of the compound microscope is often attributed to spectacle makers in the Netherlands around 1600, but surviving evidence does not establish a single inventor securely. Early compound instruments combined an objective lens near the specimen with an eyepiece near the observer. They enlarged objects considerably, yet suffered from severe color fringes, distorted images, and poor illumination. Consequently, a carefully made single lens could outperform a more elaborate compound instrument.\n\nRobert Hooke’s Micrographia, published in 1665, demonstrated the microscope’s descriptive power through detailed illustrations of insects, plant tissues, and manufactured objects. Hooke applied the word “cells” to the compartments he observed in thin slices of cork, contributing terminology that later became central to cell biology. In the 1670s, Antonie van Leeuwenhoek used small, high-quality single-lens microscopes to observe microorganisms, including protozoa and bacteria. His instruments required close positioning of the eye and specimen but achieved unusually fine detail for their period. Reports communicated to the Royal Society helped establish microscopic observation as a reproducible form of investigation.\n\nNineteenth-century improvements transformed microscopy from a craft-dependent practice into a more systematic technology. Achromatic objectives reduced chromatic aberration by combining glasses with different optical properties. Joseph Jackson Lister’s work around 1830 showed how lens combinations could reduce both chromatic and spherical aberration. Better objectives contributed to the observations behind cell theory, associated especially with Matthias Schleiden in 1838 and Theodor Schwann in 1839. Improvements in sectioning and staining were equally important: an optically sharp image was useful only when relevant structures could be exposed and distinguished.\n\nErnst Abbe’s work in the 1870s connected microscopic resolution to wavelength, diffraction, and aperture, providing a theoretical basis for objective design. His collaboration with Carl Zeiss and the glass chemist Otto Schott linked optical theory, precision manufacturing, and new glass formulations. August Köhler introduced his influential illumination arrangement in 1893. During the twentieth century, phase contrast, fluorescence, differential interference contrast, and confocal scanning greatly expanded the range of observable specimens. Digital cameras and computers subsequently made image processing and quantitative analysis integral to microscopy, while super-resolution fluorescence methods extended spatial discrimination beyond the conventional diffraction limit."
    },
    {
      "heading": "Instrument architecture and image formation",
      "text": "A simple microscope uses a single magnifying lens or lens assembly. A compound microscope uses an objective to collect light from the specimen and an eyepiece or camera system to display the resulting image. In a traditional finite-conjugate design, the objective forms a real intermediate image at a specified tube length. In an infinity-corrected design, light emerging from the objective is approximately collimated for an in-focus specimen, and a separate tube lens forms the intermediate image. The space between objective and tube lens accommodates filters, beam splitters, and other components without requiring the same changes to imaging geometry as a finite-conjugate system.\n\nThe objective largely determines resolution, light collection, working distance, and correction of aberrations. Typical markings specify magnification, numerical aperture, immersion medium, and intended coverslip thickness. Achromats correct selected chromatic and spherical aberrations; apochromats provide more extensive correction, while “plan” objectives are designed to produce a flatter image field. Correction quality matters especially in multicolor imaging, where different wavelengths must represent the same spatial position. High-magnification objectives often have short working distances, although specialized long-working-distance designs accommodate thick vessels or inaccessible surfaces.\n\nIllumination is a separate optical system rather than merely a source of brightness. In transmitted-light microscopy, a condenser directs light through the specimen into the objective. In Köhler illumination, the light source is imaged into an aperture plane rather than onto the specimen, while the field diaphragm is imaged into the specimen plane. This separation helps produce even illumination and allows independent control of the illuminated field and angular illumination range. Closing the condenser aperture can increase apparent contrast and depth of field, but excessive closure reduces resolution and can introduce conspicuous diffraction effects.\n\nInstrument geometry reflects specimen requirements. Upright microscopes place the objective above the specimen and are common for slides and many materials applications. Inverted microscopes place it below the specimen, making them useful for cells growing on the bottom of culture vessels. Stereomicroscopes provide separate viewing paths for the two eyes, yielding depth perception at relatively low magnification. Reflected-light microscopes illuminate and observe through the same objective or through associated optical paths, allowing examination of opaque metals, semiconductor devices, and rock surfaces.\n\nFor visual observation, total magnification is approximately the product of objective and eyepiece magnifications. A \\(40\\times\\) objective with a \\(10\\times\\) eyepiece therefore gives \\(400\\times\\). Camera systems instead depend on objective magnification, intermediate optics, sensor dimensions, and display scaling. Enlarging an image on a screen does not itself reveal additional specimen detail; the information must already have been transmitted by the optical system and sampled by the detector."
    },
    {
      "heading": "Resolution, diffraction, and optical performance",
      "text": "Resolution is the ability to distinguish nearby features, whereas magnification describes image enlargement. Their distinction follows from diffraction: a point emitter does not form an infinitesimal image through a finite aperture, but a distribution of intensity with a central maximum and surrounding structure. The three-dimensional response to an ideal point source is called the point-spread function. An extended object is imaged through the combined effects of this response, illumination, and detection. Increasing magnification without improving the transmitted spatial information produces empty magnification.\n\nA central objective parameter is numerical aperture, defined as\n\n\\[\\mathrm{NA}=n\\sin\\theta,\\]\n\nwhere \\(n\\) is the refractive index of the medium between specimen and objective and \\(\\theta\\) is the half-angle of the accepted light cone. A commonly used Rayleigh estimate for lateral separation is\n\n\\[d\\approx\\frac{0.61\\lambda}{\\mathrm{NA}},\\]\n\nwhere \\(\\lambda\\) is the vacuum wavelength. At \\(550\\) nanometres and an NA of \\(1.4\\), this gives approximately \\(240\\) nanometres. Abbe’s often-quoted expression, \\(d\\approx\\lambda/(2\\mathrm{NA})\\), describes a related spatial-frequency limit under particular imaging conditions. These expressions are not interchangeable universal thresholds: illumination coherence, object structure, contrast, and the criterion used to define separation all matter.\n\nImmersion media increase the accessible numerical aperture and can reduce refractive mismatch. Dry objectives operate through air and have numerical apertures below 1. Oil-immersion objectives can reach values around 1.4, while water-immersion objectives are particularly useful for aqueous specimens. High numerical aperture also improves light collection, but it generally narrows the axial range that appears sharply focused. Axial resolution is poorer than lateral resolution and scales approximately as \\(n\\lambda/\\mathrm{NA}^{2}\\), with the numerical coefficient depending on the imaging method and resolution criterion.\n\nAberrations impose additional limitations. Spherical aberration can arise when light passes through a coverslip or specimen medium different from that assumed in the objective design. It becomes especially serious when imaging deeply into tissue. Chromatic aberration can shift focus or position between colors; coma and astigmatism distort images away from ideal symmetry. Mechanical vibration, thermal drift, imperfect focus, and specimen motion can further reduce practical performance. Thus a nominal diffraction limit is not a guarantee of achieved resolution.\n\nDetector sampling must also match optical performance. Under the Nyquist–Shannon sampling theorem, a spatial frequency must be sampled at more than twice its frequency to avoid aliasing; practical microscope acquisition commonly uses several pixels across a resolvable feature. A camera with more pixels cannot recover detail excluded by the objective, while pixels that are too large can discard available detail. Resolution claims therefore require attention to the complete imaging chain rather than the objective specification alone."
    },
    {
      "heading": "Contrast in transmitted and reflected light",
      "text": "Many biological specimens absorb little visible light. An unstained cell may alter the phase of a transmitted wave without greatly changing its intensity, making it difficult to see in ordinary bright-field imaging. Contrast techniques convert differences in absorption, scattering, optical path length, polarization, or surface reflectivity into visible intensity variations. Because these mechanisms respond to different specimen properties, images of the same object can look substantially different without any change in the object itself.\n\nIn bright-field microscopy, the specimen appears against an illuminated background. Absorbing pigments and stains generate contrast by reducing transmitted intensity at selected wavelengths. This mode is widely used for prepared tissue sections and stained microorganisms. Dark-field microscopy excludes the directly transmitted illumination from the objective and collects light redirected by the specimen through scattering. Fine particles, edges, and slender organisms can then appear bright against a dark background. Dark-field visibility does not necessarily imply that the dimensions of a subresolution object can be measured accurately; detection and resolution are different capabilities.\n\nPhase-contrast microscopy, developed by Frits Zernike in the 1930s, makes transparent structures visible by modifying the relationship between undeviated and diffracted light. An annular condenser aperture and a phase plate in the objective’s rear focal plane shift and attenuate selected light components. Their interference converts phase differences into intensity differences. Optical path length depends on both refractive index and thickness, so nuclei, organelles, and cell boundaries become conspicuous without staining. Characteristic halos and shade-off artifacts, however, complicate interpretation and can limit measurements near strong boundaries.\n\nDifferential interference contrast microscopy, associated with Georges Nomarski’s work in the 1950s, uses polarized light and prisms to produce two laterally displaced beams. After traversing nearby specimen regions, the beams recombine, translating differences in optical path into intensity. The resulting image often resembles a shaded relief map, but the shading does not directly represent surface height. Contrast depends on the direction of the beam separation, and birefringent specimen holders can interfere with the method.\n\nPolarized-light microscopy exploits polarization changes caused by anisotropic materials. Between crossed polarizers, birefringent crystals can display brightness and interference colors related to orientation, thickness, and optical properties. This is important in mineral identification, polymer characterization, and the examination of ordered biological materials. Reflected-light bright-field and dark-field methods extend related contrast principles to opaque objects. In all these modes, contrast is a selective representation of an optical property, not a neutral photograph of every structure present."
    },
    {
      "heading": "Fluorescence and molecular specificity",
      "text": "Fluorescence microscopy detects light emitted after a specimen absorbs excitation light. A fluorophore usually emits at longer wavelengths than those used for excitation, allowing optical filters to separate weak emission from much stronger illumination. In a common epifluorescence arrangement, a dichroic beam splitter reflects excitation light toward the objective and transmits returning fluorescence toward the detector. Because the objective both illuminates and collects emission, the system can observe labeled structures against a comparatively dark background.\n\nFluorescent labeling connects microscopy to molecular identity. Dyes can preferentially bind DNA or accumulate in particular cellular compartments. Fluorophore-conjugated antibodies recognize selected antigens, enabling immunofluorescence imaging of proteins and other targets. Genetically encoded fluorescent proteins allow selected gene products to be followed in living systems. The demonstration of green fluorescent protein as a biological marker in 1994 was particularly influential because fluorescence could be produced through gene expression rather than solely by adding an external stain. Tags nevertheless require validation: fusion to a fluorescent protein can alter the localization, abundance, or behavior of the labeled protein.\n\nThe detected signal depends on excitation intensity, absorption probability, fluorescence quantum yield, collection efficiency, filter transmission, and detector efficiency. More illumination does not produce unlimited improvement. Fluorophores can enter non-emissive states, saturate, or undergo photobleaching, an irreversible loss of fluorescence. Excitation can also damage living specimens through photochemical reactions, including the production of reactive oxygen species. In time-lapse experiments, the relationships among illumination dose, temporal resolution, spatial detail, and specimen viability are therefore central to interpretation.\n\nMulticolor microscopy uses labels with different excitation and emission spectra to distinguish components. Spectral overlap can cause one fluorophore to appear in another detection channel, while autofluorescence from the specimen contributes additional background. Separate control specimens, spectral measurements, and computational unmixing can help distinguish these contributions. Apparent overlap between two colors also has spatial limits: colocalization within a diffraction-limited region does not establish direct molecular binding.\n\nFluorescence can report more than location. Fluorescence recovery after photobleaching follows the return of signal to a bleached region and can reveal molecular mobility and exchange. Förster resonance energy transfer measures distance-dependent energy transfer between suitable donor and acceptor fluorophores, making it sensitive to separations on the nanometre scale. Fluorescence lifetime imaging measures the temporal decay of emission after excitation and can distinguish molecular environments or energy-transfer states. Such methods turn the microscope into a spatially resolved probe of molecular behavior, provided that the physical model and labeling chemistry are appropriate."
    },
    {
      "heading": "Optical sectioning and three-dimensional imaging",
      "text": "Wide-field microscopy collects light from a substantial specimen volume. In thick fluorescent specimens, out-of-focus emission can obscure the focal plane even when individual structures would otherwise be resolvable. Confocal microscopy addresses this problem by restricting illumination and detection to corresponding focal regions. Marvin Minsky described a confocal arrangement in a patent filed in 1957; later developments in lasers, scanning, and computing made the method broadly practical. In a point-scanning instrument, a focused beam moves across the specimen and a pinhole suppresses much of the out-of-focus emission before detection.\n\nConfocal imaging creates optical sections without physically cutting the specimen. Recording successive focal planes produces a three-dimensional image stack. The pinhole size controls a trade-off between optical sectioning and detected signal; a very small pinhole rejects more light but can make acquisition slow or noisy. Spinning-disk systems scan many pinholes simultaneously, enabling faster imaging of some living specimens, although pinhole cross-talk can reduce sectioning performance in thick or strongly fluorescent samples. Confocal microscopy improves axial discrimination substantially but does not automatically provide large gains in lateral resolution.\n\nMultiphoton microscopy confines excitation through a nonlinear process. In two-photon fluorescence, a fluorophore absorbs two photons within a very short interval, usually from a tightly focused, pulsed near-infrared laser. Because excitation probability depends approximately on the square of intensity, fluorescence is concentrated near the focus. This reduces out-of-focus excitation and allows optical sectioning without a confocal detection pinhole. Near-infrared light often penetrates scattering tissue more effectively than shorter-wavelength excitation, although achievable depth still depends on absorption, scattering, labeling, and the tolerable light dose.\n\nLight-sheet microscopy illuminates a thin plane from the side while a separate objective observes fluorescence, commonly along a roughly perpendicular direction. Only a fraction of the specimen is excited at one time, reducing unnecessary exposure during volume imaging. Modern biological implementations became especially influential after selective-plane illumination microscopy was demonstrated in 2004. Light-sheet systems are widely used to follow embryonic development, image organoids, and examine large cleared tissues. Their geometry can complicate specimen mounting, and shadows or uneven illumination may require multiple views and computational fusion.\n\nThree-dimensional images remain subject to anisotropic resolution, refractive distortion, and motion between planes. A stack is not an instantaneous volume unless all relevant planes are acquired together. Deconvolution uses a measured or modeled point-spread function to estimate the underlying object from blurred data, improving contrast and sometimes the effective recovery of spatial information. Its outcome depends on noise assumptions, sampling, and model accuracy. Depth-dependent aberrations and changes in specimen refractive index can make a single, spatially invariant blur model inadequate."
    },
    {
      "heading": "Super-resolution and single-molecule localization",
      "text": "Super-resolution microscopy encompasses methods that obtain spatial information beyond the conventional far-field diffraction limit by changing the illumination, exploiting fluorophore behavior, or using additional prior information. It does not abolish diffraction. Instead, it avoids assumptions that underlie conventional resolution limits, such as the simultaneous observation of many continuously emitting, indistinguishable sources. The 2014 Nobel Prize in Chemistry recognized Eric Betzig, Stefan Hell, and William Moerner for developments in super-resolved fluorescence microscopy.\n\nStimulated emission depletion microscopy, or STED, combines an excitation focus with a depletion beam that suppresses fluorescence around its center through stimulated emission. A doughnut-shaped depletion pattern leaves a small central region able to fluoresce. Increasing depletion intensity can shrink the effective emitting region below the size of the ordinary excitation focus. In a simplified model, the attainable lateral scale follows\n\n\\[d_{\\mathrm{STED}}\\approx \\frac{\\lambda}{2\\mathrm{NA}\\sqrt{1+I/I_{\\mathrm{sat}}}},\\]\n\nwhere \\(I\\) is depletion intensity and \\(I_{\\mathrm{sat}}\\) characterizes depletion saturation. Actual performance also depends on beam quality, labeling, photobleaching, background, and specimen stability. Resolution in the tens of nanometres is achievable under suitable conditions.\n\nSingle-molecule localization methods, including PALM and STORM, separate emitters in time rather than relying on an optically smaller focus. Only a sparse subset of fluorophores is allowed to emit during each exposure, so individual diffraction patterns can be fitted to estimate molecular positions. Repeating activation or switching over many frames produces a reconstructed map. For an ideal isolated emitter, localization uncertainty decreases approximately as\n\n\\[\\sigma_{\\mathrm{loc}}\\propto\\frac{s}{\\sqrt{N}},\\]\n\nwhere \\(s\\) characterizes the image width and \\(N\\) is the detected photon count. Background, pixelation, drift, and fitting assumptions modify this relationship. Localization precision is not identical to structural resolution, which also depends on labeling density and the ability to sample neighboring features.\n\nStructured illumination microscopy projects known patterns onto the specimen and uses their interaction with fine specimen structure to shift otherwise inaccessible information into the observable frequency range. Conventional linear implementations can improve resolution by approximately a factor of two. Reconstruction requires multiple pattern positions and orientations, making specimen movement and illumination errors important sources of artifact. Other approaches, including MINFLUX, use the position-dependent response to a patterned excitation minimum to localize emitters with high precision while collecting comparatively few photons.\n\nAt nanometre scales, sample preparation and interpretation can become more limiting than the optical calculation. An antibody may position its attached dye several nanometres or more away from the target site; repeated blinking can make one molecule appear as several detections; and incomplete labeling can create false gaps. Reliable super-resolution measurements therefore distinguish localization precision, image resolution, labeling accuracy, and biological variability rather than treating a single quoted nanometre value as a complete description."
    },
    {
      "heading": "Specimen preparation and live imaging",
      "text": "Specimen preparation determines which structures are preserved, visible, and accessible to light. Thin specimens reduce scattering and out-of-focus background, while thick specimens may retain biologically important relationships. For conventional tissue microscopy, samples are commonly fixed, embedded, sectioned, and stained. Formaldehyde-based fixation cross-links molecular components, whereas alcohol-based treatments can precipitate proteins and extract lipids. Neither reproduces the living state perfectly. Fixation can alter dimensions, membrane appearance, antigen accessibility, and the distribution of soluble molecules.\n\nParaffin sections used in routine histology are often only a few micrometres thick. Hematoxylin and eosin staining provides contrasting views of nuclei and many cytoplasmic and extracellular components. Frozen sections permit faster preparation and preserve some features that may be altered by paraffin processing. Resin embedding and ultrathin or semithin sectioning serve other purposes, including high-resolution structural comparison. Stains and mounting media are part of the optical system: their absorption, refractive index, fluorescence, and chemical stability affect the resulting image.\n\nLive-cell imaging avoids many fixation artifacts but imposes environmental constraints. Temperature, pH, oxygen availability, nutrients, and osmotic conditions can affect cell behavior during observation. Incubated stages and enclosed microscope chambers help maintain defined conditions over time. Objective heating and environmental changes can cause focus drift, so long experiments often use feedback-based focus stabilization. The presence of a sharp and apparently normal cell image alone does not establish that the observed physiology remains unperturbed.\n\nThick tissues introduce refractive-index heterogeneity and scattering. Optical clearing methods reduce these obstacles by removing or modifying scattering components and matching refractive indices. Different procedures preserve fluorescent proteins, lipids, fine morphology, and molecular targets to different degrees. Some produce shrinkage or swelling that must be considered in spatial measurements. Expansion microscopy takes a deliberately different approach: biomolecules or labels are anchored to a swellable polymer network, which physically enlarges the specimen so that previously unresolved separations become accessible to conventional optics.\n\nPreparation can also determine apparent causation. A disappearing fluorescent structure may reflect bleaching rather than molecular loss; a stationary object may be immobilized by the mounting medium; and a change after adding a dye may result from the dye itself. Comparisons among live and fixed preparations, unlabeled controls, alternative labels, and independent measurement methods help identify such effects. In applications requiring repeated observation, preparation is therefore not merely a preliminary step but a controlled component of the experiment."
    },
    {
      "heading": "Quantitative imaging and scientific applications",
      "text": "Modern optical microscopy produces measurements as well as images. Common outputs include object counts, dimensions, fluorescence intensities, trajectories, spatial distributions, and rates of change. Their validity depends on calibration and an explicit relationship between image values and the quantity being inferred. A stage micrometer can establish image scale, while fluorescent standards or reference specimens can characterize intensity response, illumination uniformity, and day-to-day stability. Calibration must correspond to the actual objective, camera configuration, and acquisition settings.\n\nDigital detectors convert collected photons into electrical signals. Scientific cameras are characterized by quantum efficiency, read noise, dark current, pixel size, dynamic range, and saturation behavior. Photon arrival statistics impose shot noise, whose standard deviation scales approximately with the square root of the detected photon number. Background subtraction can remove an estimated offset but cannot remove the photon noise contributed by that background. Saturated pixels lose quantitative information, and nonlinear display adjustments can conceal weak signals or exaggerate apparent boundaries without improving the underlying data.\n\nImage analysis often begins with segmentation: separating objects or regions from their surroundings. Thresholding may work for isolated, high-contrast objects, while crowded tissues require more elaborate algorithms. Machine learning can identify cell types, segment organelles, and estimate image quality, but its results depend on training data and the conditions represented in them. A model trained on one stain, microscope, or tissue preparation may fail on another. Analysis also introduces measurement uncertainty through choices such as background estimation, object boundaries, track linking, and exclusion rules.\n\nApplications span multiple scales and disciplines. In microbiology, optical microscopy reveals microbial morphology, division, and interactions. In cell biology, fluorescent reporters follow the cytoskeleton, organelle dynamics, and signaling. Histopathology uses tissue architecture and cellular appearance as evidence in diagnosis, often supplemented by molecular stains. In geology, polarized-light examination of thin sections identifies minerals and their textural relationships. Reflected-light and interferometric methods characterize polished metals, coatings, microfabricated devices, and surface topography. These uses differ in specimen preparation and contrast mechanism, even when they share similar objectives and imaging components.\n\nReproducible quantitative work records the acquisition and processing conditions that affect interpretation: objective and immersion medium, illumination wavelength and intensity, exposure, detector settings, sampling, optical section spacing, and analysis procedures. Raw data remain distinct from contrast-adjusted illustrations and computational reconstructions. When optical measurements are combined with electron microscopy, spectroscopy, or molecular assays, the principal challenge is often registration—establishing that different measurements describe corresponding structures or states. Correlative approaches can connect the molecular specificity and live-imaging capabilities of light microscopy with structural or chemical information unavailable from optical images alone."
    }
  ],
  "links": [
    "light",
    "microscope",
    "biology",
    "optics",
    "cell",
    "bacteria",
    "royal-society",
    "diffraction",
    "numerical-aperture",
    "refractive-index",
    "nyquist-shannon-sampling-theorem",
    "scattering",
    "phase-contrast-microscopy",
    "interference",
    "polarization",
    "fluorescence-microscopy",
    "dna",
    "antibody",
    "protein",
    "confocal-microscopy",
    "super-resolution-microscopy",
    "nobel-prize-in-chemistry",
    "temperature",
    "ph",
    "calibration",
    "photon",
    "machine-learning",
    "measurement-uncertainty",
    "microbiology",
    "cytoskeleton",
    "geology"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}