Simulation-driven machine learning approach for high-speed correction of slope-dependent error in coherence scanning interferometry.

Slope-dependent error often occurs in the coherence scanning interferometry (CSI) measurement of functional engineering surfaces with complex geometries. Previous studies have shown that these errors can be corrected through the characterization and phase inversion of the instrument's three-dimensional (3D) surface transfer function. However, since CSI instrument is usually not completely shift-invariant, the 3D surface transfer function characterization and correction must be repeated for different regions of the full field of view, resulting in a long computational process and a reduction of measurement efficiency. In this work, we introduce a machine learning approach based on a deep neural network that is trainable for slope-dependent error correction in CSI. Our method leverages a deep neural network to directly learn errors characteristics from simulated surface measurements provided by a previously validated physics-based virtual CSI method. The experimental results demonstrate that the trained network is capable of correcting the surface height map with 1024 × 1024 sampling points within 0.1 seconds, covering a 178 µm field of view. The accuracy is comparable to the previous phase inversion approach while the new method is two orders of magnitude faster under the same computational condition.

Low-dose imaging denoising with one pair of noisy images.

Low-dose imaging techniques have many important applications in diverse fields, from biological engineering to materials science. Samples can be protected from phototoxicity or radiation-induced damage using low-dose illumination. However, imaging under a low-dose condition is dominated by Poisson noise and additive Gaussian noise, which seriously affects the imaging quality, such as signal-to-noise ratio, contrast, and resolution. In this work, we demonstrate a low-dose imaging denoising method that incorporates the noise statistical model into a deep neural network. One pair of noisy images is used instead of clear target labels and the parameters of the network are optimized by the noise statistical model. The proposed method is evaluated using simulation data of the optical microscope, and scanning transmission electron microscope under different low-dose illumination conditions. In order to capture two noisy measurements of the same information in a dynamic process, we built an optical microscope that is capable of capturing a pair of images with independent and identically distributed noises in one shot. A biological dynamic process under low-dose condition imaging is performed and reconstructed with the proposed method. We experimentally demonstrate that the proposed method is effective on an optical microscope, fluorescence microscope, and scanning transmission electron microscope, and show that the reconstructed images are improved in terms of signal-to-noise ratio and spatial resolution. We believe that the proposed method could be applied to a wide range of low-dose imaging systems from biological to material science.

Coherent modulation imaging using a physics-driven neural network.

Coherent modulation imaging (CMI) is a lessness diffraction imaging technique, which uses an iterative algorithm to reconstruct a complex field from a single intensity diffraction pattern. Deep learning as a powerful optimization method can be used to solve highly ill-conditioned problems, including complex field phase retrieval. In this study, a physics-driven neural network for CMI is developed, termed CMINet, to reconstruct the complex-valued object from a single diffraction pattern. The developed approach optimizes the network's weights by a customized physical-model-based loss function, instead of using any ground truth of the reconstructed object for training beforehand. Simulation experiment results show that the developed CMINet has a high reconstruction quality with less noise and robustness to physical parameters. Besides, a trained CMINet can be used to reconstruct a dynamic process with a fast speed instead of iterations frame-by-frame. The biological experiment results show that CMINet can reconstruct high-quality amplitude and phase images with more sharp details, which is practical for biological imaging applications.

Beam-pointing drift prediction in pulsed lasers by a probabilistic learning approach.

In laser systems, it is well known that beam pointing is shifted due to many un-modeled factors, such as vibrations from the hardware platform and air disturbance. In addition, beam-pointing shift also varies with laser sources as well as time, rendering the modeling of shifting errors difficult. While a few works have addressed the problem of predicting shift dynamics, several challenges still remain. Specifically, a generic approach that can be easily applied to different laser systems is highly desired. In contrast to physical modeling approaches, we aim to predict beam-pointing drift using a well-established probabilistic learning approach, i.e., the Gaussian mixture model. By exploiting sampled datapoints (collected from the laser system) comprising time and corresponding shifting errors, the joint distribution of time and shifting error can be estimated. Subsequently, Gaussian mixture regression is employed to predict the shifting error at any query time. The proposed learning scheme is verified in a pulsed laser system (1064 nm, Nd:YAG, 100 Hz), showing that the drift prediction approach achieves remarkable performances.

Adaptive aberration correction of a 5 J/6.6 ns/200 Hz solid-state Nd:YAG laser.

In this Letter, we present an adaptive aberration correction system to simultaneously compensate for aberrations and reshaping the beams. A low-order aberration corrector is adapted. In this corrector, four lenses are mounted on a motorized rail, whose positions can be obtained using a ray tracing method based on the beam parameters detected by a wavefront sensor. After automatic correction, the PV value of the wavefront is reduced from 26.47 to 1.91 µm, and the beam quality ß is improved from 18.42 to 2.86 times that of the diffraction limit. After further correction with a deformable mirror, the PV value of the wavefront is less than 0.45 µm, and the beam quality is 1.64 times that of the diffraction limit. To the best of our knowledge, this is the highest performance from such a high-power, high-pulse repetition rate Nd:YAG solid-state laser ever built.

Highly efficient and aberration-free off-plane grating spectrometer and monochromator for EUV-soft X-ray applications.

We demonstrate a novel flat-field, dual-optic imaging EUV-soft X-ray spectrometer and monochromator that attains an unprecedented throughput efficiency exceeding 60% by design, along with a superb spectral resolution of λ/Δλ > 200 accomplished without employing variable line spacing gratings. Exploiting the benefits of the conical diffraction geometry, the optical system is globally optimized in multidimensional parameter space to guarantee optimal imaging performance over a broad spectral range while maintaining circular and elliptical polarization states at the first, second, and third diffraction orders. Moreover, our analysis indicates minimal temporal dispersion, with pulse broadening confined within 80 fs tail-to-tail and an FWHM value of 29 fs, which enables ultrafast spectroscopic and pump-probe studies with femtosecond accuracy. Furthermore, the spectrometer can be effortlessly transformed into a monochromator spanning the EUV-soft X-ray spectral region using a single grating with an aberration-free spatial profile. Such capability allows coherent diffractive imaging applications to be conducted with highly monochromatic light in a broad spectral range and extended to the soft X-ray region with minimal photon loss, thus facilitating state-of-the-art imaging of intricate nano- and bio-systems, with a significantly enhanced spatiotemporal resolution, down to the nanometer-femtosecond level.