Robust projection parameter calibration in cryo-ET with L₁-norm optimization.

Area of Science:

• * Structural biology
• * Biophysics
• * Computational imaging

Background:

• * Fiducial marker-based alignment is crucial for cryo-electron tomography (cryo-ET) image reconstruction.
• * Accurate calibration of projection parameters is essential for high-resolution 3D models.
• * Existing methods struggle with noise and outliers, often assuming incorrect noise distributions.

Purpose of the Study:

• * To develop a more robust projection parameter calibration model for cryo-ET.
• * To address limitations of current methods in handling noise and outliers.
• * To improve the accuracy and reliability of cryo-ET alignment.

Main Methods:

• * Proposed a novel robust calibration model using L1-norm optimization.
• * Incorporated a Laplace noise assumption to better model real-world data.
• * Developed a fast and stable first-order optimization method with smooth approximation.

Main Results:

• * The L1-norm optimization model demonstrated superior robustness against noise and outliers.
• * The proposed method achieved higher accuracy in simulated and real-world cryo-ET data.
• * The new optimization technique improved computational speed and stability.

Conclusions:

• * The L1-norm optimization under Laplace noise provides a more effective approach for cryo-ET projection parameter calibration.
• * This method enhances the overall accuracy and reliability of cryo-ET reconstructions.
• * The developed optimization strategy offers practical advantages for cryo-ET data processing.