Centre of Rotation

The Centre of Rotation (CoR) aligns the sample’s rotation axis with the acquisition coordinate system, as shown in Fig. 11. Reconstruction assumes this alignment, so an inaccurate CoR can distort the result and invalidate later analysis.

An offset sinogram (\(d\) in Fig. 11) produces arching artefacts around object boundaries, as shown in Fig. 12. These artefacts increase with the distance from the correct CoR (\(d=0\)), so CoR estimation typically searches for the value that minimises them.

Fig. 11 The CoR offset \(d\) translates the sample coordinates \((x,y)\) into the acquisition coordinates \((s,p)\): \((s = x +- d, p = y)\).

Fig. 12 Reconstructions with different CoR values. Boundary artefacts decrease as \(d\) approaches the correct value.

CoR in HTTomo

Every reconstruction template provides a center parameter. Set it automatically (see Auto-centering) or manually (see Manual Centering).

Auto-centering

Several methods can estimate the CoR automatically. DLS commonly uses Nghia Vo’s Fourier-based sinogram method (paper), implemented by TomoPy and HTTomolibGPU. In HTTomo it is available as the find_center_vo template; see Available methods. If one automatic method fails, try another HTTomolibGPU centring method.

To use automatic centering:

1. Add the centering method before reconstruction, preferably immediately after the loader; see Group methods by data pattern.

2. Store its calculated CoR as a Side outputs.

3. Reference that output in the reconstruction method’s center parameter.

- method: find_center_vo
  module_path: httomolibgpu.recon.rotation
  parameters:
    ind: mid
    smin: -50
    smax: 50
    srad: 6
    step: 0.25
    ratio: 0.5
    drop: 20
  id: centering
  side_outputs:
    cor: centre_of_rotation
- method: FBP
  module_path: httomolibgpu.recon.algorithm
  parameters:
    center: ${{centering.side_outputs.centre_of_rotation}}
    filter_freq_cutoff: 1.1
    recon_size: null
    recon_mask_radius: null

Manual Centering

Automatic centering can fail when projection data is corrupt or incomplete, or when the sample extends beyond the detector’s field of view. In these cases, set the CoR manually. Parameter Sweeping can help identify the value.

To set it without parameter sweeping:

1. Remove or comment out the automatic centering method.

2. Replace the side-output reference in the reconstruction method’s center parameter with a numeric value.