TomoScore: A Neural Network Approach for Quality Assessment of Cellular cryo-ET

Area of Science:

• Structural Biology
• Cell Biology
• Microscopy

Background:

• Electron cryo-tomography (cryo-ET) enables label-free 3D visualization of cellular structures.
• Tomogram quality is highly variable, necessitating expert assessment for data processing.
• Different cryo-ET applications (e.g., annotation vs. averaging) have distinct quality requirements.

Purpose of the Study:

• To develop an automated tool for assessing cryo-ET tomogram quality for cellular annotation.
• To provide a quantitative measure of tomogram suitability for distinguishing subcellular features.
• To investigate the impact of electron dose on tomogram quality.

Main Methods:

• Development of a deep learning-based screening tool named TomoScore.
• Application of TomoScore to assess tomogram suitability for cellular annotation.
• Analysis of the relationship between accumulated electron dose and tomogram quality.

Main Results:

• TomoScore provides a quantitative measure for tomogram quality relevant to cellular annotation.
• The tool automates the pre-selection of tomograms, reducing manual expert involvement.
• An optimal electron dose range for cryo-ET data collection was suggested.

Conclusions:

• TomoScore effectively automates the quality assessment of cryo-ET data for cellular annotation tasks.
• The developed tool enhances efficiency in cryo-ET data processing pipelines.
• Findings provide insights into optimizing cryo-ET data acquisition parameters.