[Paper Review] Phase Retrieval with Application to Optical Imaging

Yoav Shechtman, Yonina C. Eldar+4 more|arXiv (Cornell University)|Feb 28, 2014

Advanced X-ray Imaging Techniques149 references96 citations

TL;DR

This paper provides a comprehensive review of phase retrieval in optical imaging, focusing on recovering lost phase information from intensity-only measurements using advanced algorithms. It details how Fourier-based methods, constraints like support and non-negativity, and holographic extensions enable high-resolution imaging of complex biological and nanoscale structures, even in the absence of direct phase measurement.

ABSTRACT

This review article provides a contemporary overview of phase retrieval in optical imaging, linking the relevant optical physics to the information processing methods and algorithms. Its purpose is to describe the current state of the art in this area, identify challenges, and suggest vision and areas where signal processing methods can have a large impact on optical imaging and on the world of imaging at large, with applications in a variety of fields ranging from biology and chemistry to physics and engineering.

Motivation & Objective

• To provide a contemporary review of algorithmic phase retrieval in optical imaging, linking optical physics with signal processing methods.
• To address the fundamental challenge of recovering phase information from intensity-only measurements in coherent optical systems.
• To explore the application of phase retrieval to high-resolution imaging of non-crystalline and biological specimens using X-ray and optical diffraction.
• To identify current bottlenecks in resolution, noise robustness, and real-time operation, and to outline future research directions.
• To highlight the role of signal processing in enabling atomic-scale, real-time imaging of dynamic molecular processes.

Proposed method

• Utilizes the Fourier transform relationship between near-field and far-field wave patterns in coherent optical systems.
• Applies iterative algorithms such as the Hybrid Input-Output (HIO) method to recover phase from Fourier intensity measurements.
• Incorporates prior knowledge such as object support, non-negativity, and sparsity to constrain the solution space.
• Introduces Fourier holography by embedding known reference structures (e.g., a delta function) to enhance phase retrieval stability and resolution.
• Leverages modern coherent X-ray sources (e.g., XFELs, synchrotrons) and high-harmonic generation to access sub-wavelength and attosecond-scale imaging.
• Combines physical modeling of light-matter interactions with algorithmic recovery to handle complex, dynamic systems.

Experimental results

Research questions

• RQ1 How can phase be recovered from intensity-only measurements in optical and X-ray imaging systems?
• RQ2 What constraints (e.g., support, non-negativity) are most effective in stabilizing phase retrieval for complex objects?
• RQ3 How can holographic methods improve resolution and robustness in phase retrieval without direct phase measurement?
• RQ4 What are the fundamental limits of resolution and temporal fidelity in phase retrieval for ultrafast molecular dynamics?
• RQ5 How can signal processing algorithms be adapted to handle the high bandwidth and complexity of attosecond X-ray pulses?

Key findings

• Phase retrieval enables high-resolution imaging of non-crystalline objects using only intensity measurements, bypassing the need for phase-sensitive detectors.
• The 1999 experimental demonstration of coherent diffraction imaging (CDI) of a non-periodic object marked a turning point in the revival of phase retrieval research.
• Iterative algorithms like HIO successfully reconstruct 2D images from Fourier magnitude data, as demonstrated in numerical examples with high fidelity.
• Holographic extensions, such as introducing a known reference wave, significantly improve reconstruction stability and reduce reliance on strong prior constraints.
• Current state-of-the-art imaging achieves nanoscale resolution, but atomic-level resolution remains limited by coherent X-ray flux and source availability.
• Ultrafast imaging at attosecond timescales is emerging, but challenges remain in handling the broad bandwidth and complex dynamics of such pulses.