[mitk-users] MITK segmentation

[Olfa Graa]

hello everybody,

I am actually using the MITK software to get a segmented model (of a specific organ). The problem is that when we segment an object in MITK we loose information about the original position of the object. I tried to get the original position and orientation by applying the 'translate' vector and the 'Direction Matrix' to the model. But unfortunately it doesn't work.
Please can anyone help me to understand how MITK do to initialize the position and orientation of a segmented model.
Any answer will be welcome do not hesitate if you have any idea. Thank you.

Have a nice day.

[Dinkelacker, Stefan]

Hi Olfa,

I’m not sure what you refer to when writing about segmentation models. In case of segmentation images you wouldn’t lose any information, as for example the nrrd image format preserves the origin in its header. In case of surface models most file formats do not support this extra information and as such the absolute coordinates are saved instead. When loading both into MITK, the original image and the surface model they are aligned again, even though the coordinates of the surface model are relative to the absolute origin now.

Best,
Stefan

From: Olfa Graa [mailto:***@etu.udamail.fr]
Sent: Freitag, 14. April 2017 09:31
To: mitk-***@lists.sourceforge.net
Subject: [mitk-users] MITK segmentation

hello everybody,

I am actually using the MITK software to get a segmented model (of a specific organ). The problem is that when we segment an object in MITK we loose information about the original position of the object. I tried to get the original position and orientation by applying the 'translate' vector and the 'Direction Matrix' to the model. But unfortunately it doesn't work.
Please can anyone help me to understand how MITK do to initialize the position and orientation of a segmented model.
Any answer will be welcome do not hesitate if you have any idea. Thank you.

Have a nice day.

[Olfa Graa]

Hi Stefan,

Thank you for your reply.
I'm using MRI images and I extract the segmented model as a (.ply) file. Than I transform it to an (.obj) model so I could manipulate it as a points cloud.
I wanted to know which/how transformation are applied to the segmented object, so I could find the original position.

Best,
Olfa


De: "Dinkelacker, Stefan" <***@dkfz-heidelberg.de>
À: "Olfa Graa" <***@etu.udamail.fr>, mitk-***@lists.sourceforge.net
Envoyé: Mardi 18 Avril 2017 11:46:27
Objet: RE: MITK segmentation



Hi Olfa,



I’m not sure what you refer to when writing about segmentation models. In case of segmentation images you wouldn’t lose any information, as for example the nrrd image format preserves the origin in its header. In case of surface models most file formats do not support this extra information and as such the absolute coordinates are saved instead. When loading both into MITK, the original image and the surface model they are aligned again, even though the coordinates of the surface model are relative to the absolute origin now.



Best,
Stefan




From: Olfa Graa [mailto:***@etu.udamail.fr]
Sent: Freitag, 14. April 2017 09:31
To: mitk-***@lists.sourceforge.net
Subject: [mitk-users] MITK segmentation





hello everybody,

I am actually using the MITK software to get a segmented model (of a specific organ). The problem is that when we segment an object in MITK we loose information about the original position of the object. I tried to get the original position and orientation by applying the 'translate' vector and the 'Direction Matrix' to the model. But unfortunately it doesn't work.
Please can anyone help me to understand how MITK do to initialize the position and orientation of a segmented model.
Any answer will be welcome do not hesitate if you have any idea. Thank you.

Have a nice day.

[Dinkelacker, Stefan]

From the users perspective, you can right click on your surface model in the Data Manager and click on Details to see all of the transformation stuff. From the developers perspective you can get the geometry information directly from the mitk::Surface objects via the GetGeometry() or GetTimeGeometry() methods.

Best,
Stefan

From: Olfa Graa [mailto:***@etu.udamail.fr]
Sent: Dienstag, 18. April 2017 15:52
To: Dinkelacker, Stefan
Cc: mitk-***@lists.sourceforge.net
Subject: Re: MITK segmentation

Hi Stefan,

Thank you for your reply.
I'm using MRI images and I extract the segmented model as a (.ply) file. Than I transform it to an (.obj) model so I could manipulate it as a points cloud.
I wanted to know which/how transformation are applied to the segmented object, so I could find the original position.

Best,
Olfa

________________________________
De: "Dinkelacker, Stefan" <***@dkfz-heidelberg.de<mailto:***@dkfz-heidelberg.de>>
À: "Olfa Graa" <***@etu.udamail.fr<mailto:***@etu.udamail.fr>>, mitk-***@lists.sourceforge.net<mailto:mitk-***@lists.sourceforge.net>
Envoyé: Mardi 18 Avril 2017 11:46:27
Objet: RE: MITK segmentation

Hi Olfa,

I’m not sure what you refer to when writing about segmentation models. In case of segmentation images you wouldn’t lose any information, as for example the nrrd image format preserves the origin in its header. In case of surface models most file formats do not support this extra information and as such the absolute coordinates are saved instead. When loading both into MITK, the original image and the surface model they are aligned again, even though the coordinates of the surface model are relative to the absolute origin now.

Best,
Stefan

From: Olfa Graa [mailto:***@etu.udamail.fr]
Sent: Freitag, 14. April 2017 09:31
To: mitk-***@lists.sourceforge.net<mailto:mitk-***@lists.sourceforge.net>
Subject: [mitk-users] MITK segmentation

hello everybody,

I am actually using the MITK software to get a segmented model (of a specific organ). The problem is that when we segment an object in MITK we loose information about the original position of the object. I tried to get the original position and orientation by applying the 'translate' vector and the 'Direction Matrix' to the model. But unfortunately it doesn't work.
Please can anyone help me to understand how MITK do to initialize the position and orientation of a segmented model.
Any answer will be welcome do not hesitate if you have any idea. Thank you.

Have a nice day.

[Olfa Graa]

Hi,

That's what I'm doing. In fact I get information from the Details window (I can even get information from the DICOM header). It's okay when I apply the translation from the translation vector. However It does't work for the rotation ('direction vector'). That's why I wanted to know how MITK do to get this direction vector and how to apply it to the model to get the right position.
PS : I'm working with the DTI series.

Best,
Olfa


De: "Dinkelacker, Stefan" <***@dkfz-heidelberg.de>
À: "Olfa Graa" <***@etu.udamail.fr>
Cc: mitk-***@lists.sourceforge.net
Envoyé: Mardi 18 Avril 2017 17:26:52
Objet: RE: MITK segmentation

Post by Dinkelacker, Stefan
From the users perspective, you can right click on your surface model in the Data Manager and click on Details to see all of the transformation stuff. From the developers perspective you can get the geometry information directly from the mitk::Surface objects via the GetGeometry() or GetTimeGeometry() methods.

Best,
Stefan




From: Olfa Graa [mailto:***@etu.udamail.fr]
Sent: Dienstag, 18. April 2017 15:52
To: Dinkelacker, Stefan
Cc: mitk-***@lists.sourceforge.net
Subject: Re: MITK segmentation





Hi Stefan,

Thank you for your reply.
I'm using MRI images and I extract the segmented model as a (.ply) file. Than I transform it to an (.obj) model so I could manipulate it as a points cloud.
I wanted to know which/how transformation are applied to the segmented object, so I could find the original position.





Best,
Olfa






De: "Dinkelacker, Stefan" < ***@dkfz-heidelberg.de >
À: "Olfa Graa" < ***@etu.udamail.fr >, mitk-***@lists.sourceforge.net
Envoyé: Mardi 18 Avril 2017 11:46:27
Objet: RE: MITK segmentation





Hi Olfa,



I’m not sure what you refer to when writing about segmentation models. In case of segmentation images you wouldn’t lose any information, as for example the nrrd image format preserves the origin in its header. In case of surface models most file formats do not support this extra information and as such the absolute coordinates are saved instead. When loading both into MITK, the original image and the surface model they are aligned again, even though the coordinates of the surface model are relative to the absolute origin now.



Best,
Stefan




From: Olfa Graa [ mailto:***@etu.udamail.fr ]
Sent: Freitag, 14. April 2017 09:31
To: mitk-***@lists.sourceforge.net
Subject: [mitk-users] MITK segmentation





hello everybody,

I am actually using the MITK software to get a segmented model (of a specific organ). The problem is that when we segment an object in MITK we loose information about the original position of the object. I tried to get the original position and orientation by applying the 'translate' vector and the 'Direction Matrix' to the model. But unfortunately it doesn't work.
Please can anyone help me to understand how MITK do to initialize the position and orientation of a segmented model.
Any answer will be welcome do not hesitate if you have any idea. Thank you.

Have a nice day.