Simultaneous reconstruction and segmentation with the Mumford-Shah functional for electron tomography

Area of Science:

• Structural biology
• Biophysics
• Imaging science

Background:

• Electron micrography (EM) is crucial for 3D structure determination of biological specimens.
• EM faces limitations like poor signal-to-noise, limited tilt angles, and specimen movement, leading to ill-posed reconstruction problems.
• Existing reconstruction methods can introduce artifacts.

Purpose of the Study:

• To develop an improved reconstruction method for electron micrography.
• To address limitations in 3D structure determination from EM data.
• To reduce noise and artifacts while preserving image details.

Main Methods:

• Combined an artifact reduction strategy with an iterative reconstruction algorithm.
• Utilized a Mumford-Shah model for regularization and simultaneous segmentation.
• Applied the method to simulated data (Shepp-Logan phantom) and cryo-specimen tomography.

Main Results:

• The combined method effectively regularizes ill-posed problems in EM reconstruction.
• Simultaneous segmentation and artifact smoothing were achieved.
• Noise and artifacts were significantly reduced in reconstructed images.
• Edge preservation and enhancement were demonstrated.

Conclusions:

• The developed method enhances the quality of 3D reconstructions from EM data.
• It offers a robust solution for overcoming common artifacts and noise in EM imaging.
• The technique shows promise for accurate structural analysis of macromolecular complexes and biological specimens.