help request for problems with airway segmentation

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

help request for problems with airway segmentation

brunomorra@libero.it
Hi,
 I'm a newbie of Slicer ( it's just three weeks that I'm using it)  and since I am an ENT MD ( working in Turin, Italy) interested in choking problems in paediatric age I'm trying to realize 3D-models of the upper airway at the levels of the oropharyngeal, supraglottic and glottic regions ( neck area)  and I'm interested in obtaining 3D casts of the airways in these regions ( not trachea or bronchi)

For this purpose I've started using CBCTs of these areas performed on 7 to 10 year-old kids and in the next few days I'm going to use MRIs of younger children.

After having used for a while Invivo 5 I turned to Slicer 3D because i think it's much more flexible ( but more difficult to use).

I read the tutorials on segmentation and  in the first cases I made a manual segmentation contouring with paint tool in the Editor module each slide ( and using also the Wand tool and the TracingLevelEffect)  but the process is time-consuming and the results are not vey satisfying.
 So I tried to use simple region growing segmentation and in some cases the segmentation of the airways done with Slicer works fine (and I obtain nice 3D-models) while in some other cases  the segmentation just fills all the cropped volume even if I lower both the number of iterations ( from 5 to 2 or 3) and the Multiplier ; in some other cases I obtain with this set of parameters ( lowering multiplier) models with a very rough surface ( with a lot of spikes) that it's difficult to smooth.
The fiducial is always positioned in the dark central area of the oropharynx and I think that in the cases where there is a massive "leakage" , maybe this is due to inhomogeneity of the voxel intensity in the lumen of the airways ( but I really don't know if this hypothesis is correct) 
When I've tried Airway Segmentation Module I was not able to obtain anything and when I've tried to use Robust Statistics Segmenter,  Slicer simply crashes and disappears but when I try to run it again a prompt appears saying "There are other DICOM listeners running" and asks if I want to end them or not;
 I've also  tried to use fast growcut effect  in the editor module but I've not understood the sequence of steps needed to use it.

At the present moment my step sequence is:
1) loading the images
2) open volume rendering; define the volume with ROI
3) crop volume corresponding to the volume of the airway with some surrounding tissue
4) mark a fiducial in the center of oropharynx ( where there is air)
5) apply simple region growing segmentation
6) if this works make model and save the model as a STL.

Please, could you tell me
1) if this is a correct sequence and why it seems to work in a random way even if the parameters are held identical for the different patients
2) why the other methods of segmentation are not working for me ( which mistakes am I doing ?) and 
3) how I can fix the issue and become able to use also the other segmentation methods


At the present moment I'm using a PC with 6Gb of RAM and a videocard with 1 Gb GPU but tomorrow I'm going to use  a new 32-Gb RAM machine with 6 Gb  GPU.

I'd be sincerely grateful if you can give me advice on this problem because at the present I'm working on a trial and error basis that is fairly frustrating.
Hoping to hear from you soon
Best regards and thx in advance
Bruno Morra

_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject
http://www.slicer.org/slicerWiki/index.php/Documentation/4.3/FAQ


The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.
Reply | Threaded
Open this post in threaded view
|

Re: help request for problems with airway segmentation

Kikinis, Ron,M.D.
Hi,
This sounds very reasonable. Have you tried threshold in the editor?
Ron

Sent from my mobile phone.

> On Feb 24, 2016, at 4:43 PM, "[hidden email]" <[hidden email]> wrote:
>
> Hi,
>  I'm a newbie of Slicer ( it's just three weeks that I'm using it)  and since I am an ENT MD ( working in Turin, Italy) interested in choking problems in paediatric age I'm trying to realize 3D-models of the upper airway at the levels of the oropharyngeal, supraglottic and glottic regions ( neck area)  and I'm interested in obtaining 3D casts of the airways in these regions ( not trachea or bronchi)
>
> For this purpose I've started using CBCTs of these areas performed on 7 to 10 year-old kids and in the next few days I'm going to use MRIs of younger children.
>
> After having used for a while Invivo 5 I turned to Slicer 3D because i think it's much more flexible ( but more difficult to use).
>
> I read the tutorials on segmentation and  in the first cases I made a manual segmentation contouring with paint tool in the Editor module each slide ( and using also the Wand tool and the TracingLevelEffect)  but the process is time-consuming and the results are not vey satisfying.
>  So I tried to use simple region growing segmentation and in some cases the segmentation of the airways done with Slicer works fine (and I obtain nice 3D-models) while in some other cases  the segmentation just fills all the cropped volume even if I lower both the number of iterations ( from 5 to 2 or 3) and the Multiplier ; in some other cases I obtain with this set of parameters ( lowering multiplier) models with a very rough surface ( with a lot of spikes) that it's difficult to smooth.
> The fiducial is always positioned in the dark central area of the oropharynx and I think that in the cases where there is a massive "leakage" , maybe this is due to inhomogeneity of the voxel intensity in the lumen of the airways ( but I really don't know if this hypothesis is correct)
> When I've tried Airway Segmentation Module I was not able to obtain anything and when I've tried to use Robust Statistics Segmenter,  Slicer simply crashes and disappears but when I try to run it again a prompt appears saying "There are other DICOM listeners running" and asks if I want to end them or not;
>  I've also  tried to use fast growcut effect  in the editor module but I've not understood the sequence of steps needed to use it.
>
> At the present moment my step sequence is:
> 1) loading the images
> 2) open volume rendering; define the volume with ROI
> 3) crop volume corresponding to the volume of the airway with some surrounding tissue
> 4) mark a fiducial in the center of oropharynx ( where there is air)
> 5) apply simple region growing segmentation
> 6) if this works make model and save the model as a STL.
>
> Please, could you tell me
> 1) if this is a correct sequence and why it seems to work in a random way even if the parameters are held identical for the different patients
> 2) why the other methods of segmentation are not working for me ( which mistakes am I doing ?) and
> 3) how I can fix the issue and become able to use also the other segmentation methods
>
>
> At the present moment I'm using a PC with 6Gb of RAM and a videocard with 1 Gb GPU but tomorrow I'm going to use  a new 32-Gb RAM machine with 6 Gb  GPU.
>
> I'd be sincerely grateful if you can give me advice on this problem because at the present I'm working on a trial and error basis that is fairly frustrating.
> Hoping to hear from you soon
> Best regards and thx in advance
> Bruno Morra
> _______________________________________________
> slicer-users mailing list
> [hidden email]
> http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
> To unsubscribe: send email to [hidden email] with unsubscribe as the subject
> http://www.slicer.org/slicerWiki/index.php/Documentation/4.3/FAQ
_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject
http://www.slicer.org/slicerWiki/index.php/Documentation/4.3/FAQ


The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.

Reply | Threaded
Open this post in threaded view
|

Re: help request for problems with airway segmentation

Attila Nagy
In reply to this post by brunomorra@libero.it
Hi Bruno,

just as Ron wrote, you might want to give a try to the threshold effect.
Also, maybe you could try the previous Slicer release, where there is a tool that can actually "restrict" the volume you are working on (with the threshold effect, for example) and gives very good results in a relatively short time.
You can use interpolated splines (both on the slice and between slices) to mark the ROI (it is not a ROI as per the Slicer instrumentation, I'd say :)) and then use other editor tools. This way you can work both quicker and more effectively.
You definitely should use the threshold and "use for paint" and maybe later "apply in segmented" if you still need to refine the segmentation.
I think you can find the docs for this tool in the Slicer 3.x docs under the name ModelDraw (I can't seem to find this right now...)
But here is a link where you can see how it works:
https://dl.dropboxusercontent.com/u/1686930/IMAT-2011-09-23/index.html

You can find a Slicer version with it (a 64 bit Windows one) on the Slicer download page -> Slicer3 download (at the bottom of the page) -> Snapshots -> Windows 64 bit -> and choose the version with the ...+Modeldraw string in it.

I have a video somewhere that shows a rough example on how to use it but maybe will do another one tomorrow (not today) if you need.

I know I'm old fashioned but this tool is not yet available in Slicer 4 :))

Attila

--------------------------------------------
16. 2. 24, Sze. napon Kikinis, Ron,M.D. <[hidden email]> ezt írta:

 Tárgy: Re: [slicer-users] help request for problems with airway segmentation
 Címzett:: "[hidden email]" <[hidden email]>
 Másolat: "SPL Slicer Users" <[hidden email]>
 Dátum: 2016. február. 24., Szerda, 10:03
 
 Hi,
 This
 sounds very reasonable. Have you tried threshold in the
 editor?
 Ron
 
 Sent from my mobile phone.
 
 > On Feb 24, 2016, at 4:43 PM, "[hidden email]"
 <[hidden email]>
 wrote:
 >
 > Hi,
 >  I'm a newbie of Slicer ( it's
 just three weeks that I'm using it)  and since I am an
 ENT MD ( working in Turin, Italy) interested in choking
 problems in paediatric age I'm trying to realize
 3D-models of the upper airway at the levels of the
 oropharyngeal, supraglottic and glottic regions ( neck
 area)  and I'm interested in obtaining 3D casts of the
 airways in these regions ( not trachea or bronchi)
 >
 > For this purpose
 I've started using CBCTs of these areas performed on 7
 to 10 year-old kids and in the next few days I'm going
 to use MRIs of younger children.
 >
 > After having used for a while Invivo 5 I
 turned to Slicer 3D because i think it's much more
 flexible ( but more difficult to use).
 >
 
 > I read the tutorials on segmentation
 and  in the first cases I made a manual segmentation
 contouring with paint tool in the Editor module each slide (
 and using also the Wand tool and the TracingLevelEffect) 
 but the process is time-consuming and the results are not
 vey satisfying.
 >  So I tried to use
 simple region growing segmentation and in some cases the
 segmentation of the airways done with Slicer works fine (and
 I obtain nice 3D-models) while in some other cases  the
 segmentation just fills all the cropped volume even if I
 lower both the number of iterations ( from 5 to 2 or 3) and
 the Multiplier ; in some other cases I obtain with this set
 of parameters ( lowering multiplier) models with a very
 rough surface ( with a lot of spikes) that it's
 difficult to smooth.
 > The fiducial is
 always positioned in the dark central area of the oropharynx
 and I think that in the cases where there is a massive
 "leakage" , maybe this is due to inhomogeneity of
 the voxel intensity in the lumen of the airways ( but I
 really don't know if this hypothesis is correct)
 > When I've tried Airway Segmentation
 Module I was not able to obtain anything and when I've
 tried to use Robust Statistics Segmenter,  Slicer simply
 crashes and disappears but when I try to run it again a
 prompt appears saying "There are other DICOM listeners
 running" and asks if I want to end them or not;
 >  I've also  tried to use fast
 growcut effect  in the editor module but I've not
 understood the sequence of steps needed to use it.
 >
 > At the present
 moment my step sequence is:
 > 1) loading
 the images
 > 2) open volume rendering;
 define the volume with ROI
 > 3) crop
 volume corresponding to the volume of the airway with some
 surrounding tissue
 > 4) mark a fiducial
 in the center of oropharynx ( where there is air)
 > 5) apply simple region growing
 segmentation
 > 6) if this works make
 model and save the model as a STL.
 >
 > Please, could you tell me
 > 1) if this is a correct sequence and why
 it seems to work in a random way even if the parameters are
 held identical for the different patients
 > 2) why the other methods of segmentation
 are not working for me ( which mistakes am I doing ?) and
 
 > 3) how I can fix the issue and become
 able to use also the other segmentation methods
 >
 >
 > At the present moment I'm using a PC
 with 6Gb of RAM and a videocard with 1 Gb GPU but tomorrow
 I'm going to use  a new 32-Gb RAM machine with 6 Gb 
 GPU.
 >
 > I'd be
 sincerely grateful if you can give me advice on this problem
 because at the present I'm working on a trial and error
 basis that is fairly frustrating.
 >
 Hoping to hear from you soon
 > Best
 regards and thx in advance
 > Bruno
 Morra
 >
 _______________________________________________
 > slicer-users mailing list
 > [hidden email]
 > http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
 > To unsubscribe: send email to [hidden email]
 with unsubscribe as the subject
 > http://www.slicer.org/slicerWiki/index.php/Documentation/4.3/FAQ
 _______________________________________________
 slicer-users mailing list
 [hidden email]
 http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
 To unsubscribe: send email to [hidden email]
 with unsubscribe as the subject
 http://www.slicer.org/slicerWiki/index.php/Documentation/4.3/FAQ
 
 
 The
 information in this e-mail is intended only for the person
 to whom it is
 addressed. If you believe this
 e-mail was sent to you in error and the e-mail
 contains patient information, please contact
 the Partners Compliance HelpLine at
 http://www.partners.org/complianceline .
 If the e-mail was sent to you in error
 but
 does not contain patient information, please contact the
 sender and properly
 dispose of the
 e-mail.
 
_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject
http://www.slicer.org/slicerWiki/index.php/Documentation/4.3/FAQ