[Rtk-users] problem while determining the interpolation weights in 4D ROOSTER Reconstruction
Cyril Mory
cyril.mory at creatis.insa-lyon.fr
Thu Jan 11 04:01:54 EST 2018
Sorry, I made a mistake in the attached files. Ignore
peaksOverShroud.png, I intended to send peaksOverOriginalShroud.png
instead (the shroud.mhd you sent, overlaid with the sphase.txt you sent)
so you can see how I cropped the shroud to improve the results of
rtkextractshroudsignal.
Regards
Cyril
On 11/01/2018 09:57, Cyril Mory wrote:
>
> Hi Ruoyan,
>
>
> The original sphase.txt you sent a few days earlier does not look like
> the Matlab plot you just sent. I guess you modified it in the
> meantime. It does not matter much, but for one of the images I'm
> sending, I've used your old sphase.txt.
>
>
> I am not the best person to help you with your SimpleRTK problem. If
> you're used to Matlab, let's go for that way.
>
> First, you can open your shroud image using Matlab, for example with
> the function read_mhd() provided in MITK. See
> http://mitk.org/wiki/The_Medical_Imaging_Interaction_Toolkit_(MITK) .
> Second, to crop it using Matlab, simply keep the part where the
> diaphragm motion is most visible (I removed the whole top part, wide
> 285 pixels), then save the file back to croppedShroud.mhd, then rerun
> phase extraction.
>
>
> I have used another tool than Matlab to crop the image, but it
> shouldn't matter. And I have rerun phase extraction, and generated the
> overlay of peaks over shroud image in both cases. See the attached
> pictures. Note that a few respiratory cycles have been missed by the
> algorithm. You can try playing with parameters of
> rtkextractshroudsignal, but those cycles will be hard to detect
> anyway, since they correspond to irregular breathing (it looks like
> the patient coughed). You can live with that, or modify the phase file
> manually to add those cycles: the phase is nothing more than a number
> that goes linearly from 0 to 1 between one peak and the next.
>
> I hope it helps,
> Cyril
>
> On 11/01/2018 07:54, ry.meng wrote:
>> Hi Cyril,
>>
>> The new code works fine, I still got some questions about sphase.txt
>> file.
>> When building simpleRTK, I met an error "error MSB6006:
>> “cmd.exe”exit，code 1", and I didn't find a solution. So I just plot
>> sphase.txt with Matlab and compare it with respiration signals
>> provided by raw data. They are quite different. I am not very clear
>> about how to crop the shroud, can you share the method? Thanks a lot.
>>
>>>>>> Regards
>>>>>>
>>>>>> Ruoyan Meng
>>>>>>
>>
>> On 1/10/2018 00:33，Cyril Mory<cyril.mory at creatis.insa-lyon.fr>
>> <mailto:cyril.mory at creatis.insa-lyon.fr> wrote：
>>
>>> Ruoyan,
>>>
>>>
>>> From the shroud image, the signal extraction sometimes fails to
>>> find the period of the motion of the diaphragm, which results in
>>> incorrect phase detection. The easiest way to make sure your
>>> respiratory phase has been correctly detected is to used the
>>> following python script:
>>>
>>> https://www.creatis.insa-lyon.fr/~mory/showPeaksOverShroud.py
>>>
>>>
>>> You have to save your shroud image with .mhd extension, because
>>> simpleRTK can't read .mha (I think), so modify the command line
>>> that generates it (in the 4DROOSTER wiki page, that would be
>>>
>>> rtkamsterdamshroud--path . \
>>> --regexp '.*.his' \
>>> --output shroud.mha \
>>> --unsharp 650)
>>> Change "shroud.mha" into "shroud.mhd", re-run all command lines
>>> except the ROOSTER reconstruction, and then run
>>>
>>> "showPeaksOverShroud.py shroud.mhd sphase.txt"
>>>
>>> It will generate an RGB image named "peaksOverShroud.png", which
>>> shows the shroud, and overlays the detected extrema of the
>>> phase. If your detection is wrong, you'll see it. The script
>>> requires SimpleRTK, python and a few python libs.
>>>
>>> If your peaks detection is indeed incorrect, you can try
>>> cropping the shroud to the region where the periodic signal is
>>> most visible. This is usually sufficient. If you need more help,
>>> send over your shroud.mhd and shroud.raw file.
>>>
>>> Regards,
>>> Cyril
>>>
>>> On 09/01/2018 15:47, ry.meng wrote:
>>>> Hi Cyril,
>>>>
>>>> The attachment is our projections.mhdfile and we are running
>>>> the latest RTK, example data works well.
>>>> I have another question about our sphase.txt file. As my
>>>> understanding of 4D reconstruction process, spase.txt file
>>>> should be the optimal fit respiration signals. But in our file,
>>>> it only has 6 cycles, which is impossible, it confused me.
>>>> I'm now testing your new code with our data, hope it works!
>>>> Thanks for your help.
>>>>>>
>>>>>>
>>>>>> Regards
>>>>>>
>>>>>> Ruoyan Meng
>>>>>>
>>>> On 1/9/2018 20:48，Cyril Mory<cyril.mory at creatis.insa-lyon.fr>
>>>> <mailto:cyril.mory at creatis.insa-lyon.fr> wrote：
>>>>
>>>> Hi again Ruoyan,
>>>>
>>>>
>>>> I have investigated further and found the source of the
>>>> problem: the code performs a rounding of the phase signal
>>>> to two digits, which for some values of your phase ended up
>>>> to be 1. And the rest of the code assumes that the phase
>>>> belongs to [0; 1[, i.e. there should never be a value 1.
>>>> The rounding is useful for optimization: projections with
>>>> identical phases can be forward and backprojected together,
>>>> which is faster than one-by-one, and with high-precision
>>>> phase values, there are never two equal phase values.
>>>>
>>>>
>>>> So I just fixed the bug by setting the phase to 0 when its
>>>> rounding is equal to 1, which should do the trick. The pull
>>>> request is being tested, it should be merged soon.
>>>>
>>>>
>>>> Thanks for reporting your issue.
>>>>
>>>> Regards,
>>>>
>>>> Cyril
>>>>
>>>>
>>>> On 09/01/2018 11:31, Cyril Mory wrote:
>>>>>
>>>>> Hi Ruoyan,
>>>>>
>>>>>
>>>>> I've had a look at your file sphase.txt, it looks
>>>>> perfectly fine to me. The problem must lie elsewhere. Does
>>>>> the application 4D ROOSTER work with the example data
>>>>> provided on the wiki ? If it does, then the problem lies
>>>>> in your data. In that case, please run
>>>>>
>>>>>
>>>>> rtkprojections --path D:\bd\20171116 --regexp .*.his
>>>>> -oD:\bd\20171116\projections.mhd
>>>>>
>>>>>
>>>>> and send us the "projections.mhd" file. It will contain
>>>>> only the metadata of your projections, and we'll be able
>>>>> to run the same command line as you and try to reproduce
>>>>> the problem.
>>>>>
>>>>>
>>>>> If, on the other hand, it doesn't even work with the
>>>>> example data provided on the wiki, it is a software
>>>>> problem. In that case, can you first update to the latest
>>>>> RTK (the git master branch), recompile and test again ?
>>>>>
>>>>>
>>>>> Best regards,
>>>>>
>>>>> Cyril
>>>>>
>>>>>
>>>>> On 26/12/2017 10:47, ry.meng wrote:
>>>>>> Hi
>>>>>>
>>>>>> Thank you for your reply. I have checked the value in
>>>>>> sphase.txt and don't have any value larger or equal to 5.
>>>>>> Maybe the problem lies elsewhere? Thanks.
>>>>>>
>>>>>> Regards
>>>>>>
>>>>>> Ruoyan Meng
>>>>>>
>>>>>> On 12/26/2017 17:14，Simon
>>>>>> Rit<simon.rit at creatis.insa-lyon.fr>
>>>>>> <mailto:simon.rit at creatis.insa-lyon.fr> wrote：
>>>>>>
>>>>>> Hi,
>>>>>> If you look at the code
>>>>>> <https://github.com/SimonRit/RTK/blob/master/code/rtkPhasesToInterpolationWeights.hxx#L154>,
>>>>>> you'll see that the error message corresponds to the
>>>>>> case where you have a phase which equals the number
>>>>>> of frames. This cannot be since it is 0-based. In
>>>>>> other words, with the option "--frames 5", you cannot
>>>>>> have a value larger or equal to 5 in your file
>>>>>> sphase.txt.
>>>>>> Simon
>>>>>>
>>>>>> On Mon, Dec 25, 2017 at 9:36 AM, ry.meng
>>>>>> <ry.meng at siat.ac.cn <mailto:ry.meng at siat.ac.cn>> wrote:
>>>>>>
>>>>>>
>>>>>> Hi:
>>>>>> When I tried the application of 4D ROOSTER
>>>>>> Reconstruction with my own data, it will stop for
>>>>>> the following reason: problem while determining
>>>>>> the interpolation weights. Maybe you guys have
>>>>>> met this problem before and have a solution for
>>>>>> it. Thank you very much and I hope to receive
>>>>>> your reply soon.
>>>>>>
>>>>>>
>>>>>> Happy Holidays!
>>>>>>
>>>>>>
>>>>>> Regards
>>>>>>
>>>>>> Ruoyan Meng
>>>>>>
>>>>>> ------------------------------------------------------------------------------------------
>>>>>>
>>>>>> Research Center for Medical Robotics and
>>>>>> Minimally Invasive Surgical Devices,
>>>>>> Institute of Biomedical and Health Engineering,
>>>>>> Shenzhen Institutes of Advanced Technology,
>>>>>> Chinese Academy of Sciences
>>>>>> Tel: +86-18576617767 <tel:+86%20185%207661%207767>
>>>>>> Email: ry.meng at siat.ac.cn <mailto:qs.zhu at siat.ac.cn>
>>>>>> Zip: 518055
>>>>>> Add: No. 1068 Xueyuan Avenue, Nanshan, Shenzhen,
>>>>>> China
>>>>>> <https://maps.google.com/?q=No.+1068+Xueyuan+Avenue,+Nanshan,+Shenzhen,+China&entry=gmail&source=g>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> Rtk-users mailing list
>>>>>> Rtk-users at public.kitware.com
>>>>>> <mailto:Rtk-users at public.kitware.com>
>>>>>> https://public.kitware.com/mailman/listinfo/rtk-users
>>>>>> <https://public.kitware.com/mailman/listinfo/rtk-users>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> Rtk-users mailing list
>>>>>> Rtk-users at public.kitware.com
>>>>>> https://public.kitware.com/mailman/listinfo/rtk-users
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Rtk-users mailing list
>>>>> Rtk-users at public.kitware.com
>>>>> https://public.kitware.com/mailman/listinfo/rtk-users
>>>>
>>>
>>
>
>
>
> _______________________________________________
> 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/20180111/d2e15bee/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: peaksOverOriginalShroud.png
Type: image/png
Size: 329838 bytes
Desc: not available
URL: <https://public.kitware.com/pipermail/rtk-users/attachments/20180111/d2e15bee/attachment-0001.png>
More information about the Rtk-users
mailing list