Reliable deep-learning-based phase imaging with uncertainty quantification

Abstract

Emerging deep-learning (DL)-based techniques have significant potential to revolutionize biomedical imaging. However, one outstanding challenge is the lack of reliability assessment in the DL predictions, whose errors are commonly revealed only in hindsight. Here, we propose a new Bayesian convolutional neural network (BNN)-based framework that overcomes this issue by quantifying the uncertainty of DL predictions. Foremost, we show that BNN-predicted uncertainty maps provide surrogate estimates of the true error from the network model and measurement itself. The uncertainty maps characterize imperfections often unknown in real-world applications, such as noise, model error, incomplete training data, and out-of-distribution testing data. Quantifying this uncertainty provides a per-pixel estimate of the confidence level of the DL prediction as well as the quality of the model and dataset. We demonstrate this framework in the application of large space-bandwidth product phase imaging using a physics-guided coded illumination scheme. From only five multiplexed illumination measurements, our BNN predicts gigapixel phase images in both static and dynamic biological samples with quantitative credibility assessment. Furthermore, we show that low-certainty regions can identify spatially and temporally rare biological phenomena. We believe our uncertainty learning framework is widely applicable to many DL-based biomedical imaging techniques for assessing the reliability of DL predictions.

References in corpus (17)

• What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?
• Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning
• Wide-field, high-resolution Fourier ptychographic microscopy
• Phase recovery and holographic image reconstruction using deep learning in neural networks
• Deep speckle correlation: a deep learning approach towards scalable imaging through scattering media
• Accurate Uncertainties for Deep Learning Using Calibrated Regression
• Extended depth-of-field in holographic image reconstruction using deep learning based auto-focusing and phase-recovery
• Experimental robustness of Fourier Ptychography phase retrieval algorithms
• Computational illumination for high-speed in vitro Fourier ptychographic microscopy
• Deep learning approach to Fourier ptychographic microscopy
• Efficient illumination angle self-calibration in Fourier ptychography
• High-throughput intensity diffraction tomography with a computational microscope
• Using Machine-Learning to Optimize phase contrast in a Low-Cost Cellphone Microscope
• Convolutional neural networks that teach microscopes how to image
• Optimal Physical Preprocessing for Example-Based Super-Resolution
• Coded Aperture Ptychography: Uniqueness and Reconstruction
• Optimal illumination scheme for isotropic quantitative differential phase contrast microscopy

Cited by in corpus (35)

• A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challenges
• On the use of deep learning for phase recovery
• High-speed in vitro intensity diffraction tomography
• Single-Shot 3D Widefield Fluorescence Imaging with a Computational Miniature Mesoscope
• PhaseGAN: A deep-learning phase-retrieval approach for unpaired datasets
• Deep learning in biomedical optics
• A Two-step-training Deep Learning Framework for Real-time Computational Imaging without Physics Priors
• Displacement-agnostic coherent imaging through scatter with an interpretable deep neural network
• Compensating for visibility artefacts in photoacoustic imaging with a deep learning approach providing prediction uncertainties
• Adaptive 3D descattering with a dynamic synthesis network
• Snapshot Ptychography on Array cameras
• Quantitative phase imaging (QPI) through random diffusers using a diffractive optical network
• Deep learning-based photoacoustic imaging of vascular network through thick porous media
• Deep learning phase recovery: data-driven, physics-driven, or combining both?
• Deep Learning Cell Imaging through Anderson Localizing Optical Fibre
• High space-bandwidth in quantitative phase imaging using partially spatially coherent optical coherence microscopy and deep neural network
• Ultrafast Radiographic Imaging and Tracking: An overview of instruments, methods, data, and applications
• Neural Network-Based Processing and Reconstruction of Compromised Biophotonic Image Data
• Deep Probabilistic Imaging: Uncertainty Quantification and Multi-modal Solution Characterization for Computational Imaging
• SINR: Deconvolving Circular SAS Images Using Implicit Neural Representations
• Edge effect removal in Fourier ptychographic microscopy via periodic plus smooth image decomposition
• Learned, uncertainty-driven adaptive acquisition for photon-efficient scanning microscopy
• Diffuser-based computational imaging funduscope
• High-throughput fast full-color digital pathology based on Fourier ptychographic microscopy via color transfer
• Physical model simulator-trained neural network for computational 3D phase imaging of multiple-scattering samples
• Cycle Consistency-based Uncertainty Quantification of Neural Networks in Inverse Imaging Problems
• Self-supervised physics-informed generative networks for phase retrieval from a single X-ray hologram
• Inverse scattering for reflection intensity phase microscopy
• Deep-learning-driven Reliable Single-pixel Imaging with Uncertainty Approximation
• Optical deep learning nano-profilometry
• Deep Learned Optical Multiplexing for Multi-Focal Plane Microscopy
• Blaschke Product Neural Networks (BPNN): A Physics-Infused Neural Network for Phase Retrieval of Meromorphic Functions
• Uncertainty quantification for ptychography using normalizing flows
• Mode matching in the field speckle pattern after graded-index fibers
• Spectral splitting and concentration of broadband light using neural networks