The Potential of Subsampling and Inpainting for Fast Low-Dose Cryo FIB-SEM Imaging

Area of Science:

• Materials Science
• Microscopy
• Data Science

Background:

• Cryo focused ion-beam scanning electron microscopy (FIB-SEM) tomography generates large datasets over extended periods.
• Imaging beam-sensitive materials is challenging due to beam damage and devitrification, impacting data reliability and cost.

Purpose of the Study:

• To introduce and validate subsampling and inpainting as methods for efficient and reliable data acquisition in FIB-SEM.
• To enable fast, low-dose imaging without significant loss of image quality.

Main Methods:

• Implementation of subsampling and inpainting techniques for image acquisition.
• Development of new 3D data handling methods utilizing dictionary learning and inpainting algorithms.
• Utilizing newly developed microscope control software and a data recovery algorithm.

Main Results:

• Experimental data confirm the effectiveness of subsampling and inpainting for FIB-SEM data acquisition.
• Demonstrated convenience and reliability in obtaining high-quality images with reduced imaging dose and time.
• Validation of novel 3D data processing approaches.

Conclusions:

• Subsampling and inpainting are valuable tools for improving the efficiency and reliability of FIB-SEM tomography.
• These methods mitigate challenges associated with beam-sensitive materials and lengthy acquisition times.
• The developed algorithms and software offer a practical solution for advanced microscopy data acquisition.