Fact-checked by Grok 2 weeks ago

Point spread function

The point spread function (PSF) is the three-dimensional pattern formed by an system in response to an infinitely small of , representing the fundamental unit of and characterizing how the system blurs or spreads the image of that point due to optical limitations such as and aberrations. In mathematical terms, the observed image is obtained by convolving the true object with the PSF, which encapsulates the system's . For aberration-free optical systems, the PSF is typically described by the , featuring a central bright disk surrounded by concentric rings, with the disk's radius determined by the of and the of the objective lens. Physically, the PSF arises from the wave nature of light, where rays from a point source interfere constructively and destructively after passing through the lens, resulting in a symmetrical, often hourglass-shaped distribution in three dimensions. The lateral resolution limit, known as the Rayleigh criterion, is given by $0.61 \lambda / (n \sin \alpha), where \lambda is the wavelength, n is the refractive index, and \alpha is the semi-aperture angle, while axial resolution is approximately $2 \lambda / [n (1 - \cos \alpha)]. In practice, the PSF can be measured experimentally by imaging sub-resolution fluorescent beads or point-like sources, though it varies with factors like refractive index mismatches in specimens or off-axis positions in the field of view. The plays a critical role in evaluating and enhancing the performance of imaging systems across fields such as , astronomy, and , where it directly influences and is used in techniques like to sharpen blurred images. In astronomical telescopes, such as the , the PSF describes the shape of the image from a delta-function on the detector, affected by mirror design, detector size, and source energy, enabling high-resolution studies of objects. Its full-width at half-maximum (FWHM) serves as a key metric for assessing system quality, with applications extending to computational modeling for aberration correction and image restoration.

Overview

Definition and Basic Principles

The point spread function (PSF) is defined as the two-dimensional or three-dimensional distribution resulting from an point source of light after propagation through an imaging optical system. This response captures how the system spreads or blurs the point source due to inherent physical limitations and imperfections. In essence, the PSF serves as the of the optical system, providing a complete of its blurring behavior for point-like features in the object plane. The PSF quantifies degradation effects such as , optical aberrations, and , which prevent perfect reproduction of the point source as a delta in the . For an extended object, the resulting image is mathematically described as the of the object's intensity distribution with the PSF, expressed as I(\mathbf{x}) = O(\mathbf{x}) \ast \mathrm{PSF}(\mathbf{x}), where I(\mathbf{x}) is the image intensity, O(\mathbf{x}) is the object intensity, \ast denotes , and \mathbf{x} represents spatial coordinates. This linear model holds in systems that are shift-invariant, meaning the PSF remains unchanged regardless of the point source's position in the object plane, assuming the system is linear and space-invariant. A key role of the PSF is in establishing the fundamental resolution limits of imaging systems, beyond which fine details cannot be distinguished. For instance, the Rayleigh criterion defines the minimum resolvable separation between two point sources as the distance where the central maximum of one PSF's diffraction pattern coincides with the first minimum of the other, typically yielding a resolution limit of approximately $1.22 \lambda / (2 \mathrm{NA}) for circular apertures, where \lambda is the wavelength and \mathrm{NA} is the numerical aperture. In diffraction-limited systems with a circular aperture, the PSF takes the form of the pattern—a bright central disk surrounded by concentric rings—arising from the wave nature of . This pattern encapsulates the unavoidable spread due to , with the central disk containing about 84% of the total energy.

Historical Development

The concept of the point spread function (PSF) traces its origins to early 19th-century investigations into wave and limits in imaging systems. In 1835, provided the first theoretical description of the pattern produced by a circular aperture, known as the , which represents the fundamental blurring of an ideal point source in a diffraction-limited optical system. This work laid the groundwork for understanding how spreads light beyond geometric predictions. Building on this, Lord Rayleigh's 1879 analysis of microscope resolution introduced criteria for distinguishing closely spaced points based on the overlap of Airy patterns, emphasizing the role of in limiting optical performance. These foundational studies in wave established the PSF as an implicit descriptor of image degradation, though the term itself was not yet formalized. The explicit introduction of PSF terminology occurred in the mid-20th century, as optical theory integrated linear systems analysis. In the 1940s, Pierre-Michel Duffieux pioneered Fourier optics, demonstrating that the PSF is the inverse Fourier transform of the optical transfer function (OTF), thus framing imaging as a frequency-domain process. Concurrently, in the 1930s and 1940s, Fritz Zernike and A. C. S. van Heel (later Nijboer-Zernike approach) developed aberration expansions using orthogonal polynomials, linking wavefront errors directly to aberrated PSFs as impulse responses in linear shift-invariant systems. This period marked the PSF's transition from a geometric or diffraction-only concept to a comprehensive tool in optical design, influenced by emerging systems theory. The 1950s further advanced PSF modeling through Harold H. Hopkins' wave theory of aberrations, which provided mathematical frameworks for computing PSFs under various distortions, influencing from the 1970s onward. In astronomy, the PSF gained prominence after Horace W. Babcock's 1953 proposal of , which aimed to correct atmospheric turbulence-induced broadening of stellar PSFs, enabling sharper on ground-based telescopes. Similarly, in , confocal techniques in the 1980s, as implemented by researchers like John B. Pawley and Stephen C. Webb, exploited engineered PSFs to reject out-of-focus light, enhancing axial . Modern extensions of PSF concepts emerged in the 1990s with , particularly Stefan W. Hell's 1994 stimulated emission depletion (STED) method, which depletes around the excitation PSF to shrink the effective detection volume and break the diffraction limit. This engineering of the PSF for sub-diffraction imaging has since influenced fields like nanoscopy, underscoring the evolution from passive description to active manipulation in optical systems.

Theoretical Foundations

Physical Origins

The point spread function (PSF) arises fundamentally from the wave nature of light, where even an ideal appears blurred due to when passing through an optical . According to the Huygens-Fresnel principle, every point on a acts as a source of secondary spherical wavelets that interfere constructively and destructively, leading to the spreading of the light beyond the geometric shadow. This effect sets the ultimate limit on in optical systems, qualitatively described by the angular spread θ scaling as the λ divided by the diameter D (θ ≈ λ/D), meaning smaller wavelengths or larger apertures yield sharper images. Optical aberrations further distort this ideal diffraction-limited PSF by introducing deviations in the wavefront that prevent perfect focusing. Spherical aberration occurs when rays farther from the focus at different points than those near the , causing a halo-like spread in the PSF, particularly along the . Chromatic aberration arises from the wavelength-dependent of lenses, resulting in different focal points for various colors and thus a color-fringed, broadened PSF. Astigmatism, another monochromatic aberration, produces different focal lengths in perpendicular meridional and sagittal planes, elongating the PSF into an elliptical shape off-. These aberrations degrade the symmetry and concentration of the formed by pure . In non-ideal systems, additional physical factors contribute to PSF broadening. Scattering in turbid media, such as biological tissues or atmospheric particles, redirects rays randomly, creating a diffuse around point sources that convolves with the diffraction pattern. Detector pixelation imposes a finite sampling , where the PSF is effectively convolved with the pixel response, blurring fine details below the pixel scale. Motion blur from relative movement between the object and system during exposure further smears the PSF, akin to integrating over a path of shifted point images. Notably, in regimes, the PSF differs qualitatively: for incoherent , it corresponds to the distribution, which is the squared of the of the pupil function, while for coherent illumination, the PSF is the of the pupil function, reflecting its and . In astronomical observations, atmospheric introduces a dynamic PSF source through refractive index fluctuations in the air, which warp incoming wavefronts and cause time-varying blurring beyond static aberrations. These Kolmogorov-scale eddies produce seeing disks that can exceed the diffraction limit by factors of 10 or more, with the PSF evolving rapidly over milliseconds.

Mathematical Formulation

The point spread function (PSF) serves as the of an optical imaging system to a , quantifying how the system blurs an ideal function input. The coherent point spread function is the of the pupil function P(f_x, f_y). For incoherent illumination, the intensity PSF h(x,y) is the squared magnitude of the coherent PSF. This formulation arises from the system's coherent transfer function, with the incoherent PSF obtained via the of the pupil function followed by an inverse . Image formation in the system is modeled as a linear shift-invariant process, where the observed image intensity I(x,y) is the convolution of the object intensity distribution O(x',y') with the PSF: I(x,y) = \iint O(x',y') \, h(x - x', y - y') \, dx' \, dy'. This integral represents the superposition of shifted and scaled PSF copies centered at each object point, capturing the blurring effect across the image plane. In the Fourier domain, analysis simplifies through the optical transfer function (OTF), defined as the normalized Fourier transform of the PSF: \text{OTF}(f_x, f_y) = \mathcal{F}\{ h(x,y) \}. The OTF modulates the object's spatial frequencies to yield the image spectrum: \mathcal{F}\{ I(x,y) \} = \mathcal{F}\{ O(x,y) \} \cdot \text{OTF}(f_x, f_y), enabling efficient characterization of resolution limits and frequency-dependent contrast. The derives from scalar diffraction theory under approximations such as Fresnel (near-field) or Fraunhofer (far-field), where the field at the is the of the aperture-transmitted wavefront. For high (NA) systems, the Debye integral approximates this by expanding the into spherical wave contributions, integrating over angular coordinates to yield the focused field whose magnitude squared forms the . For a circular under low-NA, paraxial conditions, the radial PSF intensity follows the Airy pattern: h(r) = I_0 \left[ \frac{2 J_1(k r \sin \theta)}{k r \sin \theta} \right]^2, where J_1 is the first-order of the first kind, k = 2\pi / \lambda is the , \lambda is the , r is the radial distance from the optic axis, and \theta is the aperture semi-angle. This arises directly from the integral of a uniformly illuminated circular . In three-dimensional imaging, such as confocal or widefield microscopy, the PSF extends axially with z-dependence, forming an elongated "hourglass" shape due to defocus effects: h(x,y,z), where the axial is typically poorer than lateral by a factor of 2–3 for high-NA objectives. The full 3D convolution becomes I(x,y,z) = \iiint O(x',y',z') \, h(x - x', y - y', z - z') \, dx' \, dy' \, dz', essential for volumetric to reassign out-of-focus light. For high-NA systems involving polarized light, scalar approximations fail, necessitating vectorial models like the Richards-Wolf formulation, which computes the components near focus via angular spectrum integrals over the aplanatic . This yields a vector PSF \mathbf{h}(x,y,z) with position-dependent and depolarization effects, reducing to the scalar Airy form at low NA but revealing longitudinal field components and tighter focal spots for certain polarizations.

Measurement and Modeling

Experimental Techniques

The primary experimental technique for characterizing the point spread function () in optical systems involves a sub-resolution , such as a pinhole or fluorescent bead, and fitting the resulting intensity profile to a theoretical model like the or Gaussian approximation. In applications, 100-nm diameter fluorescent beads are routinely employed as these point sources, as their size approximates an ideal delta function relative to the diffraction limit of high-numerical-aperture objectives. This method yields the empirical by acquiring a stack of images through focus and performing least-squares fitting to extract parameters such as (). To assess aberrations, the tilted pinhole method or star test examines the PSF's asymmetry by imaging a or artificial star defocused on either side of the focal plane, where deviations from —such as fan-like patterns for or hazy rings for —diagnose specific optical flaws. In astronomical contexts, natural guide stars or artificial guide stars serve as point sources for PSF measurement in systems, enabling calibration of distortions from atmospheric turbulence. Alternative approaches include slit or knife-edge methods, where a sharp edge is scanned across the focal plane to measure the edge spread function (ESF); the line spread function (LSF) is then obtained by of the ESF, and the two-dimensional PSF reconstructed assuming . Interferometric techniques, such as algorithms applied to defocused PSF images, allow reconstruction of the complex pupil function by iteratively propagating wavefronts between pupil and focal planes, providing detailed aberration maps without direct hardware. When direct imaging is infeasible, estimates the PSF iteratively: an initial PSF model (e.g., Gaussian) is assumed, the observed image is deconvolved to estimate the object, residuals between the reconvolved and original image guide PSF updates via optimization (e.g., maximum likelihood), and iterations continue until convergence, often incorporating regularization to handle noise. Modern hardware-based methods, like Shack-Hartmann sensors introduced in the , measure local slopes across the using a microlens array, from which the overall PSF is derived in real-time for dynamic correction in .

Computational Methods

Computational methods for modeling the point spread function (PSF) rely on numerical simulations and algorithmic estimation techniques to predict or recover PSFs from optical system parameters or observed images, bypassing the need for direct physical measurements. These approaches are essential for designing imaging systems and post-processing blurred data in fields like microscopy and astronomy. Ray-tracing simulations, based on geometrical optics, approximate PSFs in aberration-dominated scenarios by tracing light rays through the optical path and aggregating their intersections at the image plane. This method excels for systems where diffraction effects are negligible compared to aberrations, such as in wide-field telescopes. Commercial software like Zemax implements sequential and non-sequential ray tracing to compute PSFs, as demonstrated in models for the James Webb Space Telescope, where rays are propagated through aspherical lenses to generate intensity distributions. For diffraction-accurate PSF computation, wave-optics propagation methods model the full evolution. The decomposes the pupil field into plane waves via , propagates each component using a that accounts for the distance, and reconstructs the field at the to yield the PSF intensity. This technique is particularly effective for high-numerical-aperture systems, offering efficient computation through fast Fourier transforms, and has been validated for near-forward simulations in biomedical . Finite-difference time-domain (FDTD) methods solve on a discretized grid to simulate time-domain field , capturing complex and in structured media like photonic devices. FDTD is computationally intensive but provides high fidelity for tightly focused beams, enabling PSF calculation in scenarios with material inhomogeneities. PSF estimation from blurred images often employs techniques, where the PSF is recovered alongside the underlying object without prior knowledge. Variational methods formulate the problem as an optimization task minimizing an energy functional that balances data fidelity, regularization on the image and PSF, and priors like sparsity or smoothness; Bayesian frameworks iteratively update hyperparameters to jointly estimate both components. A seminal iterative for this is the Richardson-Lucy (RL) deconvolution, originally developed for object restoration but extended to blind PSF estimation by alternating updates between the object and PSF. In the blind variant, the object estimate is updated via the multiplicative rule O_{k+1}(\mathbf{x}) = O_k(\mathbf{x}) \left[ \left( \frac{I}{O_k * h_k} * h_k^T \right) (\mathbf{x}) \right], where I is the observed image, h_k is the current PSF estimate, * denotes convolution, and h_k^T is the adjoint (reversed) PSF; the PSF is then updated similarly by treating the roles reversely: h_{k+1}(\mathbf{y}) = h_k(\mathbf{y}) \left[ \left( \frac{O_{k+1}}{O_{k+1} * h_k} * O_{k+1}^T \right) (\mathbf{y}) \right], with convergence typically after 20–50 iterations depending on noise levels. This method, rooted in maximum likelihood estimation for Poisson noise, has been applied to estimate 3D PSFs from spherical bead images in tomography. In the 2020s, , particularly , has advanced blind PSF recovery by learning latent representations from training data of blurred-sharp image pairs or unsupervised priors. Convolutional neural networks (CNNs) serve as deep priors for spatially variant PSFs, iteratively refining estimates in a variational-EM framework to handle noise-blind deblurring. For instance, generative models predict PSF parameters from defocused images with Pearson correlations up to 0.99 under ideal conditions, enabling recovery in without calibration. Post-2020 developments include neural networks trained on optical simulation data to forecast PSFs from wavefront aberrations, accelerating design in . GPU-accelerated libraries facilitate efficient PSF generation and processing. The PSF Generator plugin for /Fiji simulates 3D microscope PSFs using Gibson-Lanni or Born-Wolf models, supporting vectorial for realistic computations. For accelerated involving PSF estimation, CLIJ2 integrates GPU operations into Fiji workflows, achieving up to 29-fold speedups on cloud GPUs for large-scale image restoration. These tools enable rapid prototyping of computational PSFs validated against experimental data from the mathematical formulation.

Applications

Microscopy

In optical microscopy, the point spread function (PSF) fundamentally limits , with widefield microscopy exhibiting a broader PSF that incorporates contributions from out-of-focus light, leading to significant axial and reduced in thick specimens. In , employs a pinhole to reject off-axis , effectively sharpening the PSF and improving axial by a factor of approximately 1.5 to 2 compared to widefield systems, enabling clearer imaging of three-dimensional structures at cellular scales. This distinction arises because the confocal PSF is the product of the and PSFs, concentrating the detected signal near the focal plane. Deconvolution techniques address PSF-induced blurring by computationally reversing the process, restoring high-frequency details lost in the ; in space, this involves dividing the image spectrum by the PSF's , though noise amplification necessitates regularization methods like Wiener filtering to maintain signal integrity. These approaches are particularly valuable in fluorescence microscopy, where they can enhance close to the limit without modifications, improving both lateral and axial in biological samples. Super-resolution methods engineer the PSF to surpass the Abbe diffraction limit, typically expressed as d = \frac{\lambda}{2 \mathrm{NA}}, where \lambda is the and is the ; in visible-light , the PSF's full width at half maximum (FWHM) is usually 200-300 laterally, constraining observation of subcellular features. In stimulated emission depletion () microscopy, a doughnut-shaped depletion beam modulates the effective PSF to suppress outside a central , achieving resolutions down to 20-50 by shrinking the PSF tail. Similarly, structured illumination microscopy () uses patterned illumination to shift spatial frequencies into the detectable range, effectively narrowing the PSF and doubling lateral to around 100 . A notable example is 4Pi microscopy, which employs two opposing high-NA objectives to interfere counterpropagating wavefronts, narrowing the axial PSF by a factor of nearly 7—from ~500-700 nm to ~100 nm—while providing modest lateral improvement, thus enabling isotropic resolution for volumetric imaging of fluorescently labeled structures. Since the 2010s, integration of PSF control in light-sheet microscopy has advanced live-cell imaging by selectively illuminating thin planes with customized beam profiles, minimizing and while allowing 3D to refine the elongated axial PSF for high-speed, multi-view acquisition of dynamic processes in developing embryos or neural tissues.

Astronomy

In astronomical imaging, the point spread function (PSF) is profoundly influenced by Earth's atmosphere, which causes turbulence that broadens the PSF through a phenomenon known as atmospheric seeing. This turbulence arises from variations in temperature and density in air layers, leading to random wavefront distortions that blur point sources like stars into extended images. The (FWHM) of the seeing-limited PSF typically ranges from 1 to 2 arcseconds at good ground-based observatory sites, though excellent sites like can achieve medians around 0.5 arcseconds under optimal conditions. Space-based telescopes circumvent these atmospheric effects, achieving diffraction-limited PSFs determined solely by the telescope's and . For instance, the (), with its 2.4-meter mirror, delivers PSFs with FWHM values around 0.05 arcseconds in the visible band, enabling high-resolution imaging free from seeing degradation. In contrast, ground-based telescopes rely on () systems to approach similar performance; the 10-meter Keck telescopes, for example, use to reduce the PSF FWHM to approximately 0.04 arcseconds in the near-infrared H-band under good conditions, nearing the diffraction limit. Adaptive optics corrects atmospheric distortions in real time by employing wavefront sensors to measure incoming light aberrations and deformable mirrors to adjust the telescope's optics accordingly. Wavefront sensors, often Shack-Hartmann types, detect phase variations across the pupil, while deformable mirrors with hundreds of actuators reshape the wavefront to compensate, reducing the PSF to near-diffraction-limited quality and improving Strehl ratios up to 25% or higher in the infrared. Another technique, lucky imaging, mitigates seeing by capturing thousands of short-exposure frames (typically 10-100 milliseconds) and selecting the subset with the least distortion, where atmospheric turbulence momentarily aligns to produce sharper PSFs; this method has achieved resolutions approaching the diffraction limit for small telescopes without AO. In , the PSF manifests as the synthesized beam formed by interferometric arrays, which combines signals from multiple antennas to achieve high . The Atacama Large Millimeter/submillimeter Array (), for example, produces synthesized beams with FWHM resolutions around 0.1 arcseconds in certain configurations at millimeter wavelengths, enabling detailed imaging of protoplanetary disks and star-forming regions. Post-launch characterization of the (), operational since , has confirmed its PSFs exceed pre-launch predictions, with wavefront errors below 100 nanometers RMS, yielding sharp, stable profiles crucial for coronagraphy. JWST's coronagraphs, integrated into instruments like NIRCam and , suppress starlight to reveal exoplanets, as demonstrated in mid-infrared imaging of systems like , where the PSF's high encircled energy facilitates contrast ratios necessary for detecting faint companions.

Lithography

In photolithography for semiconductor manufacturing, the aerial image is formed through the convolution of the mask pattern with the lithographic point spread function (PSF), which characterizes the blurring effect of the projection optics and is strongly influenced by the numerical aperture (NA) and illumination settings such as partial coherence. The PSF determines how sharply the projected pattern reproduces the mask features on the wafer, with higher NA values narrowing the PSF to support finer resolutions, while off-axis or annular illumination can modulate its shape to enhance contrast for specific patterns. Aberrations, such as spherical or errors in the lens system, and from scattered light within the , broaden the effective PSF, reducing image contrast and limiting the minimum achievable feature size. For instance, in 193 nm systems with NA ≈ 0.9, these effects can widen the PSF tails, constraining dense features to around 90 nm half-pitch before significant blurring occurs. , often modeled as an additional on the PSF, contributes stray energy that degrades edge definition, particularly in large fields where scatter accumulates. To mitigate PSF-induced blur, resolution enhancement techniques like (OPC) pre-distort the mask patterns, adding serifs or sub-resolution features to counteract and proximity effects during aerial image formation. OPC models explicitly account for the system's PSF to predict and compensate for linewidth variations, enabling printed features closer to the target design even under sub-wavelength conditions. The dense line resolution in is fundamentally limited by the k₁ factor in Rayleigh's criterion, given by
\text{CD} = k_1 \frac{\lambda}{\text{NA}},
where the PSF width sets the practical minimum k₁ ≈ 0.25 for conventional illumination, below which sparse features may resolve but dense patterns blur.
Extreme ultraviolet (EUV) leverages a 13.5 nm to produce a inherently narrower compared to deep ultraviolet systems, facilitating sub-7 nm logic nodes by reducing limits while maintaining NA around 0.33. suites, such as ' Sentaurus Lithography (S-Litho), have incorporated PSF-based aerial image simulations since the early , with ongoing updates for EUV-specific effects like multilayer mirror scatter to optimize process windows.

Biomedical Imaging

In biomedical , the point spread function () is particularly significant in , where the eye's monochromatic aberrations—modeled using —broaden the PSF well beyond the diffraction limit, thereby degrading . These aberrations, arising from the and , increase the root-mean-square error to approximately 1.49 µm at the fovea, resulting in PSF widths that are 5–9 times larger than the diffraction-limited value of about 0.19 arcmin for a 6.7 mm . systems mitigate this by correcting higher-order aberrations with deformable mirrors, reducing the root-mean-square error to 0.06–0.15 µm and enabling near-diffraction-limited performance in retinal imaging modalities. For instance, optical coherence tomography (AO-OCT) decouples axial and transverse resolutions to achieve isotropic 2–5 µm imaging of cellular structures , as demonstrated in volumetric retinal scans since the early 2000s. A prominent example of PSF engineering is scanning laser ophthalmoscopy (AOSLO), which corrects aberrations on both incoming and outgoing light paths to produce high-fidelity images. Introduced in 2002, AOSLO uses a 37-channel deformable mirror to narrow the PSF, boosting confocal pinhole efficiency by up to 2.5 times and resolving photoreceptors and blood flow at lateral resolutions of ~2.5 µm over depths exceeding 300 µm. This technique has transformed by allowing real-time visualization of microstructures without invasive procedures, though residual higher-order aberrations and eye motion remain challenges. In computed tomography (CT) and magnetic resonance imaging (MRI), the PSF arises from reconstruction processes and directly influences spatial resolution. CT reconstruction kernels, such as smooth (FC10), standard (FC50), and edge-enhancing (FC52), shape the PSF, with edge-enhancing kernels exhibiting high sensitivity to region-of-interest size during modulation transfer function estimation—optimal sizes range from 38–50 pixels for 0.1 mm voxels to minimize artifacts. In MRI, non-uniform k-space sampling broadens the PSF through suboptimal density compensation functions, leading to sidelobes that degrade image quality; optimization via linear matrix inequalities can suppress these to -55 dB or lower, enhancing reconstruction accuracy for clinical applications. Functional MRI further highlights PSF limitations, where blood-oxygen-level-dependent signals in gray matter exhibit a spatio-temporal PSF with a full width at half maximum of 2.34 mm at 7 T, reducing spatial specificity for brain mapping. Similarly, positron emission tomography (PET) suffers from partial volume effects that emulate PSF blurring, dispersing point-source activity across volumes due to finite resolution (typically 4–6 mm), which underestimates uptake in small lesions like tumors. Ultrasound and photoacoustic imaging rely on PSFs modulated by transducer design and tissue properties. In ultrasound, the PSF reflects the pulse-echo response to point targets, broadened by transducer geometry (e.g., array aperture and focusing) and frequency-dependent medium attenuation, which simulations show distorts axial and lateral profiles—attenuation coefficients of 0.5 dB/(cm·MHz) can widen the PSF by 20–50% over 50 mm depths. Photoacoustic tomography extends this, where back-projection reconstruction yields PSFs dependent on detection geometry: spherical arrays (e.g., 100 mm diameter with 4096 elements) approximate ideal Dirac delta functions for full 4π coverage, while planar or cylindrical setups introduce lateral broadening and artifacts due to limited angular views. Recent developments in the 2020s incorporate for PSF mitigation, such as self-supervised in clinical (OCT), which estimates blind PSFs as 2D Gaussians and performs sparse to enhance resolution by ~40% (reducing ) in , without requiring paired training data—achieving contrast-to-noise ratios up to 72 in retinal scans across 840–1310 wavelengths.

References

  1. [1]
    Microscopy Basics | The Point Spread Function - Zeiss Campus
    The ideal point spread function (PSF) is the three-dimensional diffraction pattern of light emitted from an infinitely small point source in the specimen.
  2. [2]
    [PDF] Lecture 3 Conventional and Confocal Optical Microscopies: Wave ...
    Definition: The point spread function. (PSF) describes the response of an imaging system to a point source or point object (Wikipedia, 2010). Airy Function. The ...
  3. [3]
    PSF: Point Spread Function - CIAO 4.17 - Chandra X-ray Center
    Dec 10, 2024 · Describes the shape of the image produced by a delta function (point) source on the detector, also known as Point Response Function (PRF).Missing: definition | Show results with:definition
  4. [4]
    What is a Point Spread Function? - Ansys Optics
    Introduction. The Point Spread Function (PSF) of an optical system is the irradiance distribution that results from a single point source. (A telescope forming ...
  5. [5]
    POINT SPREAD FUNCTION (PSF) - Amateur Telescope Optics
    Its mathematical description is Point Spread Function (PSF), which expresses the normalized intensity distribution of the point-source image (it should be noted ...
  6. [6]
    [PDF] Point Spread Function Workshop - Caltech
    The response of our microscope to an arbitrary source (or arbitrary object) is then the convolution of the object with the point spread function (or PSF).
  7. [7]
  8. [8]
    Calculating point spread functions: methods, pitfalls, and solutions
    Jul 15, 2024 · In fluorescence microscopy, an accurate point spread function (PSF) model, defined as the impulse response of a given optical system, is an ...
  9. [9]
    XXXI. Investigations in optics, with special reference to the ...
    Investigations in optics, with special reference to the spectroscope. Lord Rayleigh FRS. Pages 261-274 | Published online: 13 May 2009.
  10. [10]
  11. [11]
    DIFFRACTION - Amateur Telescope Optics
    According to the Huygens' principle, every wavefront point is a source of secondary spherical wavelets, which spread out in the direction of propagation ...
  12. [12]
    Resolving power
    Diffraction limits the resolution according to θ = 1.22 λ/D = y/L. Here the height of the object to be resolved is y and the distance to the object is L.
  13. [13]
    Optical Aberrations - Evident Scientific
    Spherical aberrations are very important in terms of the resolution of the lens because they affect the coincident imaging of points along the optical axis and ...
  14. [14]
    Chromatic Aberration - StatPearls - NCBI Bookshelf
    Nov 2, 2023 · Chromatic aberration is when a lens cannot focus all light wavelengths to a single point, causing color mismatches. It's also called color ...
  15. [15]
    Optical Aberrations - RP Photonics
    Optical imaging systems can exhibit various kinds of aberrations, i.e., reductions of image quality, which can be analyzed and minimized with good designs.
  16. [16]
    [PDF] Basic Wavefront Aberration Theory for Optical Metrology
    Figures 22 through 24 illustrate the normalized point spread function for different amounts of third-order spherical aberration and defocus. FIG. 22.
  17. [17]
    Learning to image and track moving objects through scattering ...
    In this study, we propose a deep learning method to decode the shape and displacement information of moving objects on the plane perpendicular to the system's ...
  18. [18]
    The point-spread function of fiber-coupled area detectors - PMC
    Sep 5, 2012 · The point-spread function (PSF) of a fiber-optic taper-coupled CCD area detector was measured over five decades of intensity using a 20 μm X-ray beam and ∼2000 ...
  19. [19]
    High-resolution dynamic inversion imaging with motion-aberrations ...
    Aug 5, 2019 · The dynamic inversion imaging adopts deep learning optical flows to reconstruct the motion point spread function, enabling removing sub-pixel ...
  20. [20]
    Coherent and Incoherent Point Spread Function - SPIE Digital Library
    The point spread function (PSF) is the image of a point object. Incoherent PSF is the squared modulus of the coherent PSF, occurring when fields are ...
  21. [21]
    THE TELESCOPE POINT SPREAD FUNCTION - IOPscience
    New observations are used to accurately define the stellar point spread function produced by atmospheric turbulence at the focus of large telescopes and to ...
  22. [22]
    AO tutorial 1: turbulence
    Point Spread Function. What happens if the telescope is not ideal? An image of a point source would not be as good as Airy function, the resolution would be ...Point Spread Function · Imaging through the... · Turbulence statistics
  23. [23]
    [PDF] Calculation of vectorial diffraction in optical systems - Zhang Lab
    While the Debye–Wolf integral is favorable for high NA systems with spherical pupil geometry, the Stratton–Chu and Luneburg integrals can be more suitable ...
  24. [24]
    Electromagnetic diffraction in optical systems, II. Structure of the ...
    An investigation is made of the structure of the electromagnetic field near the focus of an aplanatic system which images a point source.
  25. [25]
    [PDF] The JWST Point Spread Function: Calculation Methods and ... - STScI
    Jun 2, 2007 · We present a discussion of the methods that can be used to model the JWST Point Spread. Function (PSF) and illustrate some of the potential ...
  26. [26]
    The ray trace point spread function (PSF)
    The scene with a set of points is transformed to an optical image using ray trace methods based on the aspherical, 2mm lens computed in Zemax. The scene is also ...
  27. [27]
  28. [28]
    [PDF] Chapter 7 Angular Spectrum Representation
    The angular spectrum representation is a mathematical technique to describe op- tical fields in homogeneous media. Optical fields are described as a ...
  29. [29]
    [PDF] Computation of tightly-focused laser beams in the FDTD method
    Abstract: We demonstrate how a tightly-focused coherent TEMmn laser beam can be computed in the finite-difference time-domain (FDTD) method.<|separator|>
  30. [30]
    (PDF) A Variational Approach for Bayesian Blind Image Deconvolution
    Aug 6, 2025 · Two iterative algorithms that simultaneously restore the image and estimate the hyperparameters are derived, based on the application of ...
  31. [31]
    Determination of 3D PSFs From Computed Tomography ... - PubMed
    Richardson Lucy (RL) deconvolution provides a method to estimate generalized (no separability or other simplifying assumptions) 3D PSFs from spheres.
  32. [32]
    [PDF] Blind PSF estimation and methods of deconvolution optimization
    The PSF estimation includes the following steps: 1. Evaluation of the matrix ... image estimate by some iteration steps essentially. 4.4. Deblurring of ...
  33. [33]
    Learning spatially variant degradation for unsupervised blind ... - NIH
    In each iteration, we use G p as the deep PSF prior to output the estimated PSFs k , and G x as the deep image prior to obtain the recovered image result x .
  34. [34]
    GitHub - idiap/psfestimation
    Our method recovers PSF parameters from the image itself with up to a squared Pearson correlation coefficient of 0.99 in ideal conditions, while remaining ...
  35. [35]
    [PDF] Variational-EM-Based Deep Learning for Noise-Blind Image ...
    This paper proposed a deep learning based method for noise-blind image deblurring. ... A machine learning ap- proach for non-blind image deconvolution. In ...Missing: PSF
  36. [36]
    BIG • PSF Generator - Biomedical Imaging Group
    PSF Generator is a software package that allows one to generate and visualize various 3D models of a microscope PSF. The current version has more than fifteen ...Missing: GPU- accelerated libraries PyMI
  37. [37]
    [PDF] Accelerated and Reproducible Fiji for image processing using GPUs ...
    Jul 16, 2022 · This paper uses cloud GPUs with Fiji/CLIJ, achieving a 29-fold speedup, and provides an open-source methodology for casual users to utilize  ...
  38. [38]
    Any Way You Slice It—A Comparison of Confocal Microscopy ... - NIH
    However, commercial spinning-disk confocal microscope (SDCM) platforms were not prevalent until around the same time as the CLSM in the late 1980s. Early ...
  39. [39]
    [PDF] 3D Deconvolution Microscopy | Microscopist.co.uk
    This unit will review the theory of image formation and characteristics of the point spread function (PSF) based on the instrument modality and objective lens ...Missing: paper | Show results with:paper
  40. [40]
    Evaluating the resolution of conventional optical microscopes ...
    Oct 20, 2023 · In optical microscope systems, the mechanism of light-intensity diffusion is generally expressed as a point spread function (PSF). Therefore, ...Missing: history | Show results with:history
  41. [41]
    Fluorescence microscopy beyond the diffraction limit - ScienceDirect
    The dimension of the PSF can be described by its full-width-half-maximum (FWHM), which has a dimension of typically half the wavelength or around 200–300 nm in ...
  42. [42]
    Super-resolution STED microscopy in live brain tissue - ScienceDirect
    To achieve this, the STED beam PSF is engineered into a doughnut shape of a high intensity ring with a 0-intensity center and is superimposed on the Gaussian ...
  43. [43]
    A guide to super-resolution fluorescence microscopy
    Jul 19, 2010 · These new super-resolution technologies are either based on tailored illumination, nonlinear fluorophore responses, or the precise localization of single ...
  44. [44]
    Superresolution References | 4Pi Microscopy - Zeiss Campus
    Furthermore, the point-spread function of a 4Pi microscope is almost 1.5-fold sharper in the lateral dimension and 7-fold sharper in the axial direction ...
  45. [45]
    Deconvolution of light sheet microscopy recordings | Scientific Reports
    Nov 26, 2019 · We developed a deconvolution software for light sheet microscopy that uses a theoretical point spread function, which we derived from a model of image ...
  46. [46]
    astronomical seeing, part 1: the nature of turbulence - Handprint.com
    Seeing that is below a FWHM of 1 arcsecond is considered excellent, and seeing that is above 5 arcsecond is considered poor. (Note that FWHM is based on the ...
  47. [47]
    Atmospheric and facility seeing on Mauna Kea, Hawaii
    From this analysis, I conclude that the atmospheric seeing at the summit of Mauna Kea is characterized by (FWHM) = 0.50 ± 0.05 (1 r) arc sec and that the ...
  48. [48]
    [PDF] lucky imaging: diffraction-limited astronomy - UNC Physics
    May 23, 2006 · The 2.5m Hubble Space Telescope (HST) has the same diffraction-limited resolution as the. 2.5m Nordic Optical Telescope. The Advanced Camera ...
  49. [49]
    Performance of the W.M. Keck Observatory Natural Guide Star ...
    ... (FWHM) went from 0.6 to 0.04 arcsec at H-band (1.65 micrometer wavelength), with a Strehl ratio of 25%. The AO system became an officially scheduled Keck ...
  50. [50]
    Astronomical adaptive optics: a review | PhotoniX | Full Text
    May 1, 2024 · Since the concept of adaptive optics(AO) was proposed in 1953, AO has become an indispensable technology for large aperture ground-based ...Missing: 1950s | Show results with:1950s<|separator|>
  51. [51]
    Lucky imaging: high angular resolution imaging in the visible from ...
    A Point Spread Function (PSF) guide star is selected as a reference to the turbulence induced blurring of each frame. 2. The guide star image in each frame is ...
  52. [52]
    ALMA Basics — ALMA Science Portal at NRAO
    ... resolutions range from 20 mas at 230 GHz to 43 mas at 110 GHz. These numbers refer to the FWHM of the synthesized beam (point spread function), which is the ...
  53. [53]
    [PDF] Characterization of JWST Science Performance from Commissioning
    Jul 12, 2022 · The optics are better aligned, the point spread function is sharper with higher encircled energy, and the optical performance is more time- ...
  54. [54]
    Understanding JWST Image Quality: Wavefront and PSF Modeling
    Nov 8, 2024 · JWST users can access archived wavefront maps and utilize STScI tools to visualize and trend JWST wavefront error, and to create high-fidelity ...
  55. [55]
    A temperate super-Jupiter imaged with JWST in the mid-infrared
    Here we report JWST coronagraphic images which reveal a giant exoplanet that ... The exoplanet detection highlights the power of using indirect evidence to target ...
  56. [56]
    Fast inverse lithography based on a model-driven block stacking ...
    By squaring and then summing the weighted convolution results of the mask and the optical kernels, the aerial image intensity can be approximated as follows:.
  57. [57]
    EUV computational lithography using accelerated topographic mask ...
    Apr 4, 2019 · In this paper, we apply a fast modeling approach to EUV light diffraction on topographic masks, which is based on fully rigorous topographic ...Missing: PSF | Show results with:PSF
  58. [58]
    Flare effect of different shape of illumination apertures in 193-nm ...
    Aug 9, 2025 · Flare has been important variable to obtain good CD control in the resolution limited lithography area such as sub-90 nm node.
  59. [59]
    Comparison of techniques to measure the point spread function due ...
    FVC needs the within-die flare level estimated by convolving the Point Spread Function due to scatter (PSFsc) with the mask layout. Thus, accurate knowledge of ...
  60. [60]
    Focus blur model to enhance lithography model for optical proximity ...
    Oct 17, 2008 · We developed several methodologies to model the laser chromatic aberration and vertical stage vibration in OPC (Optical Proximity Correction) ...Missing: PSF | Show results with:PSF
  61. [61]
    S-Litho: Predictive Lithography Simulation | Synopsys
    S-Litho EUV enables process optimization and development for the most advanced nodes. It's ready for all high-NA EUV specific simulation applications, allowing ...Missing: PSF | Show results with:PSF