Filters not implemented for FBP

[carterbox]

According to the docs, filters are not implemented for FBP. However, TomoPy does not raise an error when you try to pass filter_name or filter_par with algorithm='fbp'. This would explain why, in my tests, choosing various filters for FBP did not change the reconstruction quality.

This definitely needs to be fixed in two ways:

1. In the short term, an error should be raised when filtered are provided with FBP because they aren't applied.
2. In the long term, filters with FBP should be implemented because FBP requires filters. Otherwise, it's just BP.

[mahayusuf]

@carterbox - Just wanted to confirm whether FBP is actually only BP in tomopy? I thought "filters are not implemented" in the warning meant that there are no "additional filters" implemented on FBP because it is already filtered. But I would really like to clarify this point. Please let me know.

[carterbox]

TomoPy does not apply any filters to your data when using the fbp method, so yes, fbp is actually just bp. You need to filter your data before handing it off to TomoPy if you want fbp.

[mahayusuf]

Thank you!!

My reconstructed data has radial gradients due to truncation artifacts I got in ROI-specific uCT scan. So I want to implement my own filter before I reconstruct using BP. So its actually great that filter and BP component of FBP are separated in tomopy!

Question: can I implement my own smooth padding filter and then use tomopy's BP for reconstruction? Is BP compatible with filters other than the standard used in FBP?

[carterbox]

Can I implement my own smooth padding filter and then use tomopy's BP for reconstruction?

You can perform arbitrary preprocessing of your data as you choose.

Is BP compatible with filters other than the standard used in FBP?

The list of filters implemented in TomoPy is not exhaustive. There may be others that have been proposed. You will need to search the relevant scientific literature.