Fast, non-iterative algorithm for quantitative integration of X-ray differential phase-contrast images.

Lorenzo Massimi,Ian Buchanan,Marco Endrizzi,Alessandro Olivo,Alberto Astolfo

doi:10.1364/oe.405755

Abstract

X-ray phase contrast imaging is gaining importance as an imaging tool. However, it is common for X-ray phase detection techniques to be sensitive to the derivatives of the phase. Therefore, the integration of differential phase images is a fundamental step both to access quantitative pixel content and for further analysis such as segmentation. The integration of noisy data leads to artefacts with a severe impact on image quality and on its quantitative content. In this work, an integration method based on the Wiener filter is presented and tested using simulated and real data obtained with the edge illumination differential X-ray phase imaging method. The method is shown to provide high image quality while preserving the quantitative pixel content of the integrated image. In addition, it requires a short computational time making it suitable for large datasets.

Highlights

Differential interference contrast is an established microscopy technique for the visualization of unstained biological specimens with poor transmission or reflection contrast to visible light [1]
The phase from unidirectional differential phase contrast (DPC) images is retrieved by direct integration of each image row
We present a non-iterative integration method for unidirectional X-ray DPC images, based on an image model initially developed for differential interference contrast microscopy with visible light [20,21]

Summary

Introduction

Differential interference contrast is an established microscopy technique for the visualization of unstained biological specimens with poor transmission or reflection contrast to visible light [1]. While the retrieval of the phase from its second derivative is widely investigated for propagation-based phase imaging [13,14], gratings-based and edge illumination imaging provide access to the first derivative of the phase, in one or two directions depending on the experimental setup [15,16]. Noise in the DPC image and inevitable limitations in the sampling frequency of the phase signal, leads to the formation of streak artifacts, which severely affect the quality and the quantitative content of the integrated image. We present a non-iterative integration method for unidirectional X-ray DPC images, based on an image model initially developed for differential interference contrast microscopy with visible light [20,21]. We found that the method outperforms direct integration while providing a reliable quantitative estimation of image values if used on DPC images with a sufficiently high signal-to-noise ratio (SNR). We used the method to integrate an experimental DPC edge illumination image of a complex specimen

Objectives

Methods

Results

Conclusion