Classification and assessment of retrieved electron density maps in coherent X-ray diffraction imaging using...

Area of Science:

• Materials Science
• Biological Science
• Crystallography
• Data Analysis

Background:

• Coherent X-ray diffraction imaging (CXDI) visualizes non-crystalline particles using phase-retrieval (PR) algorithms.
• PR in CXDI faces challenges due to noise and missing data, unlike experimental phase determination in crystallography.
• Objective criteria are needed to assess PR accuracy beyond convergence monitoring.

Purpose of the Study:

• To develop an objective data analysis scheme (ASURA) for selecting accurate electron density maps in CXDI.
• To enhance structural analysis of non-crystalline particles by improving phase retrieval accuracy.
• To provide a robust method for evaluating the quality of reconstructed electron density maps.

Main Methods:

• ASURA utilizes principal component analysis and k-means clustering on a set of retrieved electron density maps.
• Maps are generated from 1000 different random seeds for a given diffraction pattern.
• The scheme objectively selects the most probable maps for structural analysis in two stages: shape estimation and fine structure determination.

Main Results:

• ASURA successfully selected accurate and probable electron density maps from diffraction data.
• The scheme was validated using data from metal particle aggregates and biological specimens at the XFEL facility SACLA.
• The method provides objective criteria for assessing the quality of reconstructed electron density maps.

Conclusions:

• The ASURA scheme offers a reliable method for objective assessment and selection of electron density maps in CXDI.
• This approach addresses limitations in phase retrieval, enabling more accurate structural analysis of non-crystalline materials and biological samples.
• ASURA enhances the utility of CXDI for advanced structural investigations at facilities like SACLA.