[vtkusers] Hounsfield Units : strange result (Volume rendering)

wassim_belhadj at topnet.tn wassim_belhadj at topnet.tn
Wed Feb 18 05:57:01 EST 2009


Hi, 

THANK YOU VERY MUCH FOR YOUR EXPLANATION.

Should I use ITK for this segmentation (region growing algorithm ).

Can you help me? 

How do I proceed (ITK-VTK)?

Have you an example of 3D segmentation (region growing algorithm)?

Regards, Wassim

On Tue, 17 Feb 2009 22:36:32 +1100, Kevin Osborn <planck at netspeed.com.au>
wrote:
> By definition the Hounsfield value of air is -1000. Water has a value of 
> 0 HU and all other CT densities (or strictly speaking electron 
> densities) represent a linear scale around these two values. In practice 
> the densities don't fit into a narrow Hounsfield range due to variations 
> in calibration of the xray detectors, noise within CT systems and so 
> called partial volume artefact where some image voxels contain both air 
> and soft tissue and therefore have a density intermediate between air 
> (-1000 HU) and soft tissue (approx 40 HU). To set an appropriate range 
> to render lung, you will always therefore render the air/skin interface 
> (and also stomach and bowel gas if the scan includes the upper abdomen). 
> As Frederic Perez said, a region growing algorithm is your best bet.
> 
> Regards, Kevin
> 
> 
> wassim_belhadj at topnet.tn wrote:
>> Hi,
>>
>> Thank you for your reply. 
>>
>> According to Hounsfield Units, the pixel values that form the air is
>> between -1000 <--------> -995.
>>
>> I want to display ONLY the lungs (not skin ...) so I set my function
>> transfer like this:
>>
>> opacityFun->AddPoint(-3024,      0.0);
>> opacityFun->AddPoint(-878.674,   0.0);
>> opacityFun->AddPoint(-864.466,   1.0);
>> opacityFun->AddPoint(-594.524,   1.0);
>> opacityFun->AddPoint(-395.62,    1.0);
>> opacityFun->AddPoint(3071,       0.0);
>>
>> Hounsfield Units for lungs are   -950 <--------> -550 
>>
>> I Attached a NEW screenshot that illustrates the problem.
>>
>> I vary the values of the transfer function but no result.I did not
manage
>> to remove the part that surrounds the lungs.
>>
>> Best regards,
>> BELHADJ wassim
>>
>>
>>
>>
>>
>> On Tue, 17 Feb 2009 18:06:18 +1100, Kevin Osborn
<planck at netspeed.com.au>
>> wrote:
>>   
>>> Don't forget air surrounds the patient. What your image shows is the 
>>> air/skin interface, not the skin itself. As the lung is mostly air,
your
>>>
>>> Hounsfield range between -600 and -400 not only gives lung but also 
>>> includes air around the patient.
>>>
>>>
>>> wassim_belhadj at topnet.tn wrote:
>>>



More information about the vtkusers mailing list