[Rtk-users] R: Trying to reconstruct with a 180 degrees circular projection geometry

[Simon Rit]

You can check https://doi.org/10.1088/0031-9155/47/14/311 which is not
implemented in RTK (yet). But only one half of the field-of-view can be
reconstructed exactly if you rate 180°. For the rest, there will be missing
data.
With RTK, iterative reconstruction should achieve a decent result but will
necessarily be limited by the missing data. I recommend the conjugate
gradient algorithm which will require adjusting the regularization
parameter Gamma and the number of iterations.

On Fri, Dec 11, 2020 at 11:26 AM Clément Mirabel <clement.mirabel at gmail.com>
wrote:

> Hi,
>
> Thank you both for your answer.
> I managed to have a good reconstruction with 200° since my cone angle is
> about 20°. I didn't mention in my original email that I was already using
> Parker weightings, but as you mentioned I needed higher angles. Thank you
> for your help.
> As a more general question, the machine I am using to acquire data can
> only rotate 180° around the patient but is still able to reconstruct (I
> don't have access to the projections, it generates a 3d volume as output).
> Are you aware of any algorithm that would be able to do so?
>
> Clément
>
> Le ven. 11 déc. 2020 à 10:03, Simon Rit <simon.rit at creatis.insa-lyon.fr>
> a écrit :
>
>> Hi Clément,
>> To complement Gabriele's response and explains what goes on. FDK requires
>> a 360° acquisition. RTK will calculate the angular gap between the current
>> projection and the two neighbors. With your configuration, two projections
>> take an angular weight (the delta when discretising the backprojection
>> integral) of about 90°.
>> You can do a short scan with Parker weightings but, as stated by
>> Gabriele, that necessitates at least 180° + the cone angle.
>> Simon
>>
>> On Fri, Dec 11, 2020 at 9:23 AM <gabriele.belotti.bergamo at gmail.com>
>> wrote:
>>
>>> Hi Clément,
>>>
>>> It seems likely that your geometry is not 180+fan-angle complete and
>>> therefore cannot lead to a proper reconstruction. From your attached
>>> geometry, you sweep from 0 to 178° (180-increment).
>>>
>>> My suggestion would be to try some intermediate angles, so to recreate a
>>> complete ShortScan. I suggest you to try at least a 200/220 ° arc to check
>>> on your simulation pipeline.
>>>
>>> I hope this is helpful,
>>> Gabriele
>>>
>>>
>>>
>>> *Da:* Rtk-users <rtk-users-bounces at public.kitware.com> *Per conto di *Clément
>>> Mirabel
>>> *Inviato:* venerdì 11 dicembre 2020 09:11
>>> *A:* rtk-users at public.kitware.com
>>> *Oggetto:* [Rtk-users] Trying to reconstruct with a 180 degrees
>>> circular projection geometry
>>>
>>>
>>>
>>> Hi,
>>>
>>>
>>>
>>> I have been trying to set the geometry in
>>> ThreeDCircularProjectionGeometry with values between 0 and 180 degrees, to
>>> reproduce the motion of some acquisition systems, but reconstruction is far
>>> from the expected result when the gantry angle does not reach max value of
>>> (360-increment). (Attached is the geometry file generated)
>>>
>>> My workflow is the following :
>>>
>>>    - Apply JosephForwardProjectionImageFilter on a 3D Volume
>>>    - Try to reconstruct using FDKConeBeamReconstructionFilter
>>>
>>> Although, the result from JosephForwardProjectionImageFilter describes a
>>> 180 degrees rotation, the output of the reconstruction seems to display
>>> rays from opposite direction but not from rotated directions
>>>
>>>
>>>
>>> Here is a view of what is supposed to be an axial slice:
>>>
>>>
>>>
>>> Thanks for your help!
>>>
>>>
>>>
>>> Clément
>>>
>>>
>>> _______________________________________________
>>> Rtk-users mailing list
>>> Rtk-users at public.kitware.com
>>> https://public.kitware.com/mailman/listinfo/rtk-users
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://public.kitware.com/pipermail/rtk-users/attachments/20201211/d76e138a/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 69568 bytes
Desc: not available
URL: <https://public.kitware.com/pipermail/rtk-users/attachments/20201211/d76e138a/attachment-0001.png>