Physics-aware machine learning (PAML) has proven to be successful in re- trieving object in a low-photon imaging condition. In this study, we take an information theoretic approach to study maximum amount of information (i.e., the imaging capacity) that can be reliably retrieved under photon-starved imaging condition. We formulate a sim- ple problem to derive the imaging capacity using Monte Carlo approach: small integrated circuit objects (made out of two different type of materials) imaged with monochromatic x-rays. The simple problem is useful in that the entropies for priors and measurements can be explicitly computed. It is a simple version of the imaging system proposed in Ref. [3]. The constructed problem will help us understand the limit on photon budget, information gain associated with varying number of scan angles, and optimal PAML algorithm that can approach the theoretic imaging capacity.
This paper aims to examine the processes associated with failure of the cornea and other collagenous tissues during photoablation. Two different constitutive models are applied to simulate a series of laser deposition experiments into porcine reticular dermis (1), a biological tissue similar to the cornea in composition and photoablation characteristics. The first of our constitutive models, DFRACT, is a physically motivated, micromechanical model based on the nucleation and growth of spherical voids (2). The second is a relatively simple model that allows the material to vaporize and thermally soften. The simulation results reproduce the prominent features observed experimentally thereby shedding a new light on the operative mechanisms during photoablation. The good qualitative agreement between the simulated stress histories and the stress histories measured during the experiments also demonstrates the effectiveness of micromechanical damage and failure modeling as a viable tool for optimizing existing laser surgery procedures and designing new ones.
X-ray tomography is a non-destructive imaging technique that reveals the interior of an object from its projections at different angles. Under sparse-view and low-photon sampling, regularization priors are required to retrieve a high-fidelity reconstruction. Recently, deep learning has been used in X-ray tomography. The prior learned from training data replaces the general-purpose priors in iterative algorithms, achieving high-quality reconstructions with a neural network. Previous studies typically assume the noise statistics of testing data is acquired a priori from training data, leaving the network susceptible to a change in the noise characteristics under practical imaging conditions. In this work, we propose a noise-resilient deep-reconstruction algorithm for X-ray tomography. By training the network with regularized reconstructions from a conventional algorithm, the learned prior shows strong noise resilience without the need for additional training with noisy examples, and allows us to obtain acceptable reconstructions with fewer photons in testing data. The advantages of our framework may further enable low-photon tomographic imaging where long acquisition times limit the ability to acquire a large training set.
The formation of zonal flows from inhomogeneous drift-wave (DW) turbulence is often described using statistical theories derived within the quasilinear approximation. However, this approximation neglects wave--wave collisions. Hence, some important effects such as the Batchelor--Kraichnan inverse-energy cascade are not captured within this approach. Here we derive a wave kinetic equation that includes a DW collision operator in the presence of zonal flows. Our derivation makes use of the Weyl calculus, the quasinormal statistical closure, and the geometrical-optics approximation. The obtained model conserves both the total enstrophy and energy of the system. The derived DW collision operator breaks down at the Rayleigh--Kuo threshold. This threshold is missed by homogeneous-turbulence theory but expected from a full-wave quasilinear analysis. In the future, this theory might help better understand the interactions between drift waves and zonal flows, including the validity domain of the quasilinear approximation that is commonly used in literature.
The exposure of human skin to near-infrared radiation is numerically simulated using coupled laser, thermal transport and mass transport numerical models. The computer model LATIS is applied in both one-dimensional and two-dimensional geometries. Zones within the skin model are comprised of a topical solder, epidermis, dermis, and fatty tissue. Each skin zone is assigned initial optical, thermal and water density properties consistent with values listed in the literature. The optical properties of each zone (i.e. scattering, absorption and anisotropy coefficients) are modeled as a kinetic function of the temperature. Finally, the water content in each zone is computed from water diffusion where water losses are accounted for by evaporative losses at the air-solder interface. The simulation results show that the inclusion of water transport and evaporative losses in the model are necessary to match experimental observations. Dynamic temperature and damage distributions are presented for the skin simulations.
A major challenge faced by the oil industry during the appraisal of a hydrocarbon discovery and the development of potential development plans is the estimation of uncertainty emanating from a diverse ranges of sources.
The computer code LATIS is used to simulate midplane and backplane spallation resulting from short pulsed laser absorption. A 1D planar geometry is simulated with an exponential laser absorption profile. The laser pulse length is assumed to be much shorter than the sound transit time across the laser absorption length. The boundary conditions are a fixed front plane and free backplane and a free front plane and a fixed midplane. The NBS/NRC equation of state for water is used with a self-consistent yet empirical material strength and failure model. The failure model includes the effects of void nucleation, growth and coalescence. Definite signatures of the nucleation and coalescence thresholds are found in the back surface motion for backplane spallation.
The exposure of human skin to near-infrared radiation is numerically simulated using coupled laser, thermal transport and mass transport numerical models. The computer model LATIS is applied in both one-dimensional and two-dimensional geometries. Zones within the skin model are comprised of a topical solder, epidermis, dermis, and fatty tissue. Each skin zone is assigned initial optical, thermal and water density properties consistent with values listed in the literature. The optical properties of each zone (i.e. scattering, absorption and anisotropy coefficients) are modeled as a kinetic function of the temperature. Finally, the water content in each zone is computed from water diffusion where water losses are accounted for by evaporative losses at the air-solder interface. The simulation results show that the inclusion of water transport and evaporative losses in the model are necessary to match experimental observations. Dynamic temperature and damage distributions are presented for the skin simulations.
The validity of an extended Rayleigh model for laser generated bubbles in soft tissue is examined. This model includes surface tension, viscosity, a realistic water equation of state, material strength and failure, stress wave emission and linear growth of interface instabilities. It is compared to detailed dynamic simulations using the computer program LATIS. These simulations include stress wave propagation, a realistic water equation of state, material strength and failure, and viscosity. The extended Rayleigh model and the detailed dynamic simulations are compared using a 1D spherical geometry with a bubble in the center and using a 2D cylindrical geometry of a laser fiber immersed in water with a bubble formed at the end of the fiber. Studies are done to test the validity of the material strength and failure, stress wave emission, and the interface instability terms in the extended Rayleigh model. The resulting bubble radii, material damage radii, the emitted stress wave energies, and the size of the interface distortions are compared. Conclusions are made on the validity of the extended Rayleigh model and on possible improvements to this model. The purpose of this study is to investigate the use of the extended Rayleigh model as a substitute for the detailed dynamic simulations when only limited information is needed. It is also meant to benchmark the detailed dynamic simulations when only limited information is needed. It is also meant to benchmark the detailed dynamic simulations and highlight the relevant physics. It is shown that the extended Rayleigh model executes over 300 times faster on a computer than the detailed dynamic simulations.
We describe algorithms for automating the process of picking seismic events in pre-stack migrated gathers. The approach uses supervised learning and statistical classification algorithms along with advanced signal-image processing algorithms. We train a probabilistic neural network (PNN) for pixel classification using event times and offsets (ground truth information) picked manually by expert interpreters. The key to success is in using effective features that capture the important behavior of the measured signals. We use a variety of features calculated in a local neighborhood about the pixel under analysis. Feature selection algorithms are used to ensure that we use only the features that maximize class separability. The novelty of the work lies in (a) the use of pre-stack migrated gathers rather than stacked data, (b) the use of two-dimensional statistical and wavelet features, and (c) the use of a PNN for classification. 8 refs., 3 figs
