Closed dyang37 closed 1 year ago
Tested
demo_3D_shepp_logan.py
anddemo_mace3D_fast.py
, both ran as expected.This pull request changes one line in
NSI_load_scans_and_params
which is only used bydemo_nsi_preprocess.py
, it does not affect anything else -> seems safe.
Thanks Marco!
When you have time, could you please also help me test demo_nsi_preprocess.py
?
The change in this PR is specific to the geometry calculation of NSI datasets. Therefore only the script demo_nsi_preprocess.py
will get affected.
Thanks!
OK. I started running it now, it looks like it will take a while. Will let you know once it's finished.
Ok. Ran successfully in a bit over an hour.
Thanks Marco! The run time is as expected, given the size of the reconstruction problem (480x384x384). Regards, Diyu
From: Marco Bommarito @.> Sent: Friday, March 24, 2023 4:08 PM To: cabouman/mbircone @.> Cc: Yang, Diyu @.>; Author @.> Subject: Re: [cabouman/mbircone] detector row offset calculation (PR #124)
---- External Email: Use caution with attachments, links, or sharing data ----
Ok. Ran successfully in a bit over an hour.
— Reply to this email directly, view it on GitHubhttps://github.com/cabouman/mbircone/pull/124#issuecomment-1483349868, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AB2Q7OESHRPIGBPBASQ7FF3W5X5LDANCNFSM6AAAAAAWG3QFMA. You are receiving this because you authored the thread.Message ID: @.***>
OK, since Marco tested this, and it takes an hour to run, and it is only one line of code, I'm accepting the PR.
This pull request fixes the det_row_offset calculation in NSI preprocess function. The fix is only one line of code :-)
Results of
demo_nsi_preprocess.py
before & after the fix:before (det_row_offset=0):
after (det_row_offset=1.15):
The results are visually similar before & after because the row offset is small.
Also tested the code with the tin-filtered dataset, which has a larger row_offset=16.76. It yields a clear reconstruction without misalignment artifacts.