[Insight-users] about initial level set

[Zachary Pincus]

Did you set the filter's IsoContourValue? This tells the filter where  
the boundary is in the initial level set. If you don't set it, it  
defaults to zero, which is bad for an binary image where either the  
inside or the outside would be zero. (This is all covered in the ITK  
software guide.)

Also, the ITK convention is that the "inside" of a shape is defined by  
values less that the IsoContourValue. So if your mask is zero outside  
of the shape and 255 inside of the shape, you'd need to negate/invert  
the image so that the inside values are less than the outside values.  
Then, with a mask that has (say) inside with values of zero and outside  
with values of one, you'd set the IsoContourValue to 0.5.

Luis knows this system better than I, and will probably have some  
additional comments/corrections about the above.

Of course, you still might have better luck converting the binary  
outline to a signed distance map; this might help the stability of the  
subsequent computations. You would use either the  
SignedDanielssonDistanceMapImageFilter (note: requires the "outside"  
value of the binary mask to be zero), or the  
ApproximateSignedDistanceMapImageFilter (faster, less accurate, and  
allows arbitrary inside and outside values).

Zach Pincus

Department of Biochemistry and Program in Biomedical Informatics
Stanford University School of Medicine



On Jun 28, 2005, at 11:12 PM, Yixun Liu wrote:

> Hi,
>
> I did as you said. However, the result is wrong.
>
> I use the ShapeDetectionLevelSetImageFilter. It's first input is the  
> segmented binary image and it's second input is a feature image,which   
> is the output of the sigmoid filter. The result is obvious wrong.  I  
> look for the document about this filter. It says that the first input  
> should be a initial level set, which is different from a binary image.  
> So, in my opinion, some filters should be used to convert the bianry  
> image to a distance map.
>
> Best regards,
>
> Yixun Liu
>
> ----- Original Message -----
> From: "Luis Ibanez" <luis.ibanez at kitware.com>
> To: "Yixun Liu" <yxliu at fudan.edu.cn>
> Cc: <insight-users at itk.org>
> Sent: Thursday, June 16, 2005 9:45 PM
> Subject: Re: [Insight-users] about initial level set
>
>
>>
>> Yixun,
>>
>>
>> Don't need to do anything particular.
>>
>> You can feed your binary image directly as input level set.
>>
>>
>> Regards,
>>
>>
>>    Luis
>>
>>
>> ----------------
>> Yixun Liu wrote:
>>
>>> Hi,
>>>
>>> I have a  segmented image represented by binary image. I want to use
>>> level set to refine the segmentation based on the segmented image.  
>>> How
>>> to generate a initial level set based on this binary image?
>>>
>>> Best regards,
>>>
>>> Yixun Liu
>>>
>>>
>>> --------------------------------------------------------------------- 
>>> ---
>>>
>>> _______________________________________________
>>> 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
>