[Vv] Errors with clitkExtractLung

Rômulo Pinho romulopinho at yahoo.com.br
Wed Oct 23 23:38:08 CEST 2013


Ok, Bilal, thanks for the update.
It's hard for me to check it now, because I'd have to download/compile 
the entire VV project plus ITK, VTK, etc.
In any case, I still suggest you to play with the parameters of 
clitkExtracLung related to the extraction of the trachea. You can also 
post the output of --verboseRG on the list.
Best,
Rômulo

On 23/10/2013 19:20, Bilal Tahir wrote:
> Hi Rômulo,
>
> Thanks for your reply. Actually, Vivien had already converted those 
> images from matlab format to mhd. I double checked them and he did 
> indeed account for the signed/unsigned HU related issues. Let me know 
> and I could put one of the images up on dropbox. I'm looking at their 
> breath-hold copd data - not 4dct.
>
> Best wishes,
>
> Bilal
>
>
>
> On 23 October 2013 21:41, Rômulo Pinho <romulopinho at yahoo.com.br 
> <mailto:romulopinho at yahoo.com.br>> wrote:
>
>     Hello, Bilal,
>
>     I'd have to verify this (maybe Vivien can also comment on that),
>     but I think the dir-lab data are not in HU, i.e., the grey values
>     are all positive, from 0 to 4096. The algorithms in
>     clitkExtractLung apply some thresholding operations that depend on
>     a data set in HU, roughly from -1024 to 3072.
>
>     To cope with it, you can either play with the parameters
>     seedPreProcessingThreshold and upperThresholdForTrachea in
>     clitkExtratLung (depending on TYPE)
>     "upperThresholdForTrachea"    -    "0: Initial upper threshold for
>     trachea"  default="-900"
>     "seedPreProcessingThreshold"   -  "1: Threshold for the
>     pre-processing step of the algorithm" default="-400"
>
>     or, to be on the safe side, shift your grey values prior to
>     launching clitkExtractLung. You can do it by creating an image
>     with -1024 everywhere (clitkCreateImage does the job) and then
>     applying the shift by adding the two images with clitkImageArithm.
>
>     In any case, using --verboseRG might also give some extra insights.
>
>     I hope this helps!
>
>     Best,
>     Rômulo
>
>
>     On 23/10/2013 14:15, Bilal Tahir wrote:
>>     Thanks Vivien,
>>
>>     I've got passed the .afdb stage. However I'm getting errors at
>>     the 'search for trachea' stage for both the original and LOLA11
>>     algorithms. The dataset is the inspiration scans for patient 1 of
>>     the dir-lab copd data.
>>
>>     1) LOLA 11:
>>
>>     [md1bat at testnode02 mhd]$ clitkExtractLung --verboseStep 1 -i
>>     copd1_00.mhd -a default.afdb -o 00_lungs.mhd --type=1
>>     Step 1 -- Copy and crop input image to 'patient' extends
>>     Step 2 -- Set background to initial image
>>     Step 3 -- Remove Air
>>     Step 4 -- Search for the trachea
>>
>>     itk::InvalidRequestedRegionError (0x2bca660)
>>     Location: "unknown"
>>     File:
>>     /home/md1bat/Source/itk/Modules/Core/Common/src/itkDataObject.cxx
>>     Line: 411
>>     Description: Requested region is (at least partially) outside the
>>     largest possible region.
>>
>>     2) Original (type=0)
>>
>>     [md1bat at testnode02 mhd]$ clitkExtractLung --verboseStep 1 -i
>>     copd1_00.mhd -a default.afdb -o 00_lungs.mhd --type=0
>>     Step 1 -- Copy and crop input image to 'patient' extends
>>     Step 2 -- Set background to initial image
>>     Step 3 -- Remove Air
>>     Step 4 -- Search for the trachea
>>            The volume of the trachea (4035.53 cc) seems not correct.
>>     I skip some slices (0) and restart to find seeds.
>>            The volume of the trachea (4037.01 cc) seems not correct.
>>     I skip some slices (5) and restart to find seeds.
>>            The volume of the trachea (4036.86 cc) seems not correct.
>>     I skip some slices (10) and restart to find seeds.
>>            The volume of the trachea (4036.39 cc) seems not correct.
>>     I skip some slices (15) and restart to find seeds.
>>            The volume of the trachea (4036.07 cc) seems not correct.
>>     I skip some slices (20) and restart to find seeds.
>>            The volume of the trachea (4035.31 cc) seems not correct.
>>     I skip some slices (25) and restart to find seeds.
>>            The volume of the trachea (4035.54 cc) seems not correct.
>>     I skip some slices (30) and restart to find seeds.
>>            The volume of the trachea (4035.49 cc) seems not correct.
>>     I skip some slices (35) and restart to find seeds.
>>            The volume of the trachea (4035.39 cc) seems not correct.
>>     I skip some slices (40) and restart to find seeds.
>>            The volume of the trachea (4035.57 cc) seems not correct.
>>     I skip some slices (45) and restart to find seeds.
>>            The volume of the trachea (4035.69 cc) seems not correct.
>>     I skip some slices (50) and restart to find seeds.
>>            The volume of the trachea (4036.09 cc) seems not correct.
>>     I skip some slices (55) and restart to find seeds.
>>            The volume of the trachea (4036.34 cc) seems not correct.
>>     I skip some slices (60) and restart to find seeds.
>>     2 : Number of slices to skip to find trachea too high = 65
>>
>>     itk::ExceptionObject (0x377f970)
>>     Location: "Unknown"
>>     File: /home/md1bat/Source/vv/segmentation/clitkExtractLungFilter.txx
>>     Line: 266
>>     Description: No seeds for trachea... Aborting.
>>
>>     3) Manually setting a higher number of slices than 65
>>
>>     [md1bat at testnode02 mhd]$ clitkExtractLung --verbose 1 -i
>>     copd1_00.mhd -a default.afdb -o 00_lungs.mhd --skipslices=90
>>     Input is 3D short.
>>     Reading [copd1_00.mhd] ...
>>     2 : Number of slices to skip to find trachea too high = 95
>>
>>     itk::ExceptionObject (0x3a379d0)
>>     Location: "Unknown"
>>     File: /home/md1bat/Source/vv/segmentation/clitkExtractLungFilter.txx
>>     Line: 266
>>     Description: No seeds for trachea... Aborting.
>>
>>     As you can see, I keep failing at the search trachea stage even
>>     by trying various different options for slices to skip for the
>>     trachea. Please let me know if you have any suggestions that can
>>     help.
>>
>>     Best wishes,
>>
>>     Bilal
>>
>>
>>     On 23 October 2013 16:51, Vivien Delmon
>>     <vivien.delmon at creatis.insa-lyon.fr
>>     <mailto:vivien.delmon at creatis.insa-lyon.fr>> wrote:
>>
>>         On 10/18/2013 12:08 AM, Bilal Tahir wrote:
>>
>>             Dear VV community,
>>
>>             ...
>>
>>
>>             Also I would like to know if the LOLA11 paper below is
>>             the correct
>>             reference if I decide to use your method for publications?
>>
>>             http://lola11.com/Results/displayApproved?result=20110708160218_86_CREATIS_CLB_Lyon_graphcut&file=20110708160218_86_CREATIS_CLB_Lyon_graphcut_lola11.pdf
>>
>>
>>         Part of the work presented in this paper have been
>>         implemented in clitkExtractLung but you need to explicitly
>>         ask the tool to use it :
>>
>>         Step 2 : find trachea:
>>           --type=INT trachea finding algorithm (0: original;
>>         1:LOLA11) (default=`0')
>>
>>         The second part about graphcut labeling were not included in
>>         this tool since we never really need to separate lungs.
>>
>>         Last time I tried to activate --type=1 it failed, I will have
>>         a look at it one day.
>>
>>         As far as I know there is no other paper that describe what
>>         we have implemented in this tool, so this paper should be the
>>         one to cite.
>>
>>         -- 
>>         Vivien Delmon
>>         PhD student at CREATIS, Lyon
>>         Supported by ANRT and ELEKTA
>>         http://www.creatis.insa-lyon.fr/rio
>>
>>         _______________________________________________
>>         vv mailing list
>>         vv at creatis.insa-lyon.fr <mailto:vv at creatis.insa-lyon.fr>
>>         http://www.creatis.insa-lyon.fr/mailman/listinfo/vv
>>
>>
>>
>>
>>     -- 
>>     Bilal Tahir,
>>     James Morrison Researcher in Radiotherapy Imaging,
>>     Department of Clinical Oncology,
>>     University of Sheffield
>>
>>
>>     _______________________________________________
>>     vv mailing list
>>     vv at creatis.insa-lyon.fr  <mailto:vv at creatis.insa-lyon.fr>
>>     http://www.creatis.insa-lyon.fr/mailman/listinfo/vv
>
>
>
>
> -- 
> Bilal Tahir,
> James Morrison Researcher in Radiotherapy Imaging,
> Department of Clinical Oncology,
> University of Sheffield

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.creatis.insa-lyon.fr/pipermail/vv/attachments/20131023/8311d508/attachment-0001.html>


More information about the vv mailing list