[Insight-users] Kideny Segmentation

[Luis Ibanez]

Hi Emiliano,

Thanks for posting the screen shots of the gamma
camera images that you are trying to segment.

The information on these images doesn't seem to
be enough for describing the anatomical shape
of the Kidney.

Would you have access to images of the same
patient acquired in another modality ?

For example CT ?

If so, then you could extract the Kidney from
that other modality and then use it as a model
for registering against the gamma camera images.

Do you have access to a 3D volume of images from
the gamma camera ?

or are you only dealing with a 2D projection of the
image (like X-rays) and its changes over time ?

The matching process will be different depending
on whether you are searching for an object in a
projection image or an object embedded in a 3D volume.


Please let us know more details about the gamma
camera images that you posted.


     Thanks



       Luis


--------------------------
Emiliano Beronich wrote:
> Thanks for your answer, Luis. I've posted some images.
> 
> http://www.veccsa.com/images/thumbs.png
> 
> This one presents intensity irregularities inside the kidneys,
> http://www.veccsa.com/images/0001602.png
> 
> The irregularities inside the kidney change the gradient and add a 
> potential edge difficult to discriminate without other considerations.
> 
> Diffuse edges,
> http://www.veccsa.com/images/0002201.png
> 
> Different intensities between left and right kidney,
> http://www.veccsa.com/images/0003101.png
> 
> The final target is to get a contour of the kidney. I hope this images 
> allows you to get a better idea of the problem. I'm reading about Model 
> based registration.
> 
> Regards,
> Emiliano
> 
> 
> 
> Luis Ibanez escribió:
> 
>>
>> Hi Emiliano,
>>
>> Can you post in a web site a couple of screen shots of your images ?
>>
>> That will help a lot to figure out the degree of irregularities
>> in your images.
>>
>> If you are having to deal simultaneously with:
>>
>>   1) Different intensities between left and right kidney
>>   2) Intensity irregularities inside the kidneys
>>   3) Diffuse edges
>>
>>
>> Then you probably may want to use model-based segmentation.
>>
>> In this approach you create a geometrical model of a kidney
>> (e.g. by segmenting a typical kidney from another modality
>> such as CT or MR where the kidney anatomy will be better
>> defined).  Then you register this geometrical model against
>> the gamma camera image that you have.
>>
>> You will find a description of this method at the end of the
>> Image Registration chapter of the ITK Software Guide:
>>
>>        http://www.itk.org/ItkSoftwareGuide.pdf
>>
>>
>> As any other segmentation method, this approach has some
>> advantages and disadvantages, but, given the constrains
>> that you listed in your email, it seems to be more appropriate
>> than methods such as level sets, watersheds and region growing.
>>
>>
>>
>> That being said...
>> We should first take a look at your images. Maybe the factors
>> that you describe are not as accentuated as we are assuming
>> here.
>>
>>
>>
>>     Regards,
>>
>>
>>
>>         Luis
>>
>>
>> ------------------------
>> Emiliano Beronich wrote:
>>
>>> Hello,
>>>
>>> I write in regard to a technical problem with segmenting kidneys from 
>>> gamma cammera images. I am basically trying to segment a kidney and I 
>>> am facing problems such as different intensity regions between right 
>>> and left kidney, irregularities inside the kidney region and diffuse 
>>> edge of the region by a slow trasition.
>>> I would like to know which strategy is recommended to deal with these 
>>> issues and which algorithm you consider better for this application.
>>>
>>>
>>> Regards,
>>>
>>>     Emiliano
>>> _______________________________________________
>>> Insight-users mailing list
>>> Insight-users at itk.org
>>> http://www.itk.org/mailman/listinfo/insight-users
>>>
>>>
>>
>>
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
> 
>