Efficient and accurate inversion of multiple scattering with deep learning

Abstract

Image reconstruction under multiple light scattering is crucial in a number of applications such as diffraction tomography. The reconstruction problem is often formulated as a nonconvex optimization, where a nonlinear measurement model is used to account for multiple scattering and regularization is used to enforce prior constraints on the object. In this paper, we propose a powerful alternative to this optimization-based view of image reconstruction by designing and training a deep convolutional neural network that can invert multiple scattered measurements to produce a high-quality image of the refractive index. Our results on both simulated and experimental datasets show that the proposed approach is substantially faster and achieves higher imaging quality compared to the state-of-the-art methods based on optimization.

References in corpus (5)

• Deep Convolutional Neural Network for Inverse Problems in Imaging
• Fast Image Recovery Using Variable Splitting and Constrained Optimization
• Learning optimal nonlinearities for iterative thresholding algorithms
• Efficient Inversion of Multiple-Scattering Model for Optical Diffraction Tomography
• A Recursive Born Approach to Nonlinear Inverse Scattering

Cited by in corpus (33)

• Physics-informed neural networks for inverse problems in nano-optics and metamaterials
• Deep speckle correlation: a deep learning approach towards scalable imaging through scattering media
• Low Photon Count Phase Retrieval Using Deep Learning
• RARE: Image Reconstruction using Deep Priors Learned without Ground Truth
• Noise2Inverse: Self-supervised deep convolutional denoising for tomography
• Deep Learning Diffuse Optical Tomography
• Physics-informed neural networks for imaging and parameter retrieval of photonic nanostructures from near-field data
• High-Resolution Limited-Angle Phase Tomography of Dense Layered Objects Using Deep Neural Networks
• SIMBA: Scalable Inversion in Optical Tomography using Deep Denoising Priors
• SOM-Net: Unrolling the Subspace-based Optimization for Solving Full-wave Inverse Scattering Problems
• Adaptive 3D descattering with a dynamic synthesis network
• Ghost imaging based on Y-net: a dynamic coding and conjugate-decoding approach
• SGD-Net: Efficient Model-Based Deep Learning with Theoretical Guarantees
• ANN-assisted CoSaMP Algorithm for Linear Electromagnetic Imaging of Spatially Sparse Domains
• Holographic particle localization under multiple scattering
• Limited-angle tomographic reconstruction of dense layered objects by dynamical machine learning
• Optimal Physical Preprocessing for Example-Based Super-Resolution
• Linear-Model-inspired Neural Network for Electromagnetic Inverse Scattering
• CoIL: Coordinate-based Internal Learning for Imaging Inverse Problems
• Estimation of moisture content distribution in porous foam using microwave tomography with neural networks
• Deep Learning for Biomedical Image Reconstruction: A Survey
• Multiscale Physics-Informed Neural Networks for the Inverse Design of Hyperuniform Optical Materials
• Role of Multiple Scattering in Single Particle Perturbations in Absorbing Random Media
• Noise-robust latent vector reconstruction in ptychography using deep generative models
• Revisit Geophysical Imaging in A New View of Physics-informed Generative Adversarial Learning
• Model-free estimation of the Cramér-Rao bound for deep-learning microscopy in complex media
• Joint Reconstruction and Calibration using Regularization by Denoising
• Inverse scattering for reflection intensity phase microscopy
• Deep learning using a biophysical model for Robust and Accelerated Reconstruction (RoAR) of quantitative and artifact-free R2* images
• Modeling Deep Learning Based Privacy Attacks on Physical Mail
• Generative Deep Learning Model for a Multi-level Nano-Optic Broadband Power Splitter
• Composition-Aware Spectroscopic Tomography
• Programming multi-level quantum gates in disordered computing reservoirs via machine learning and TensorFlow