[vtkusers] Hounsfield Units of Human body part

[Chasan KIOUTSOUKMOUSTAFA]

Wes,
Yes, 650 HU is much better for all datasets for a starting point and I see that it is nearly impossible to find optimal value without human interaction. 

Chasan KIOUTSOUKMOUSTAFA

--- On Thu, 8/26/10, Wes Turner <wes.turner at kitware.com> wrote:

From: Wes Turner <wes.turner at kitware.com>
Subject: Re: [vtkusers] Hounsfield Units of Human body part
To: "Chasan KIOUTSOUKMOUSTAFA" <chasank at yahoo.com>
Cc: "vtk mailList" <vtkusers at vtk.org>
Date: Thursday, August 26, 2010, 9:28 PM

Chasan,
Marching Cubes looks for the boundary where the voxel value goes from below the threshold value to above the threshold value.  CT is a noisy modality and you should not expect that all the bone voxels will be at the optimal bone value.  Assuming that 1250 is the average value for bone, then one half of the bone voxels will be at or below 1250.  So setting the Marching Cubes threshold to that level will give a very poor segmentation.


You should set the Marching Cubes threshold at somewhere between the value outside of the bone vs the value of the bone.   Based on your numbers, I would probably go with around 650 as a starting point and then adjust up or down based on image characteristics and the quality of the output segmentation until I reach a suitable value.  The noise level of the image, patient age, surrounding material (i.e. in vivo or ex vivo), and age of the bones (for ex vivo) will all have an impact on the determining the proper threshold.


- Wes

On Thu, Aug 26, 2010 at 2:13 PM, Chasan KIOUTSOUKMOUSTAFA <chasank at yahoo.com> wrote:


I'm using vtkDICOMImageReader class for reading the images, if i give the exact HU values for example;

boneExtractor->SetInputConnection(reader->GetOutputPort());


boneExtractor->SetValue(0, HUofBone); // HUofBone is 1250

it does not give optimal solution...
boneExtractor is vtkMarchingCubes object


Chasan KIOUTSOUKMOUSTAFA

--- On Thu, 8/26/10, John Drescher <drescherjm at gmail.com> wrote:



From: John Drescher <drescherjm at gmail.com>
Subject: Re: [vtkusers] Hounsfield Units of Human body part


To: "Chasan KIOUTSOUKMOUSTAFA" <chasank at yahoo.com>
Cc: "vtk mailList" <vtkusers at vtk.org>


Date: Thursday, August 26, 2010, 9:06 PM

On Thu, Aug 26,
 2010 at 1:59 PM, Chasan KIOUTSOUKMOUSTAFA
<chasank at yahoo.com> wrote:
>
> HUofAir = (double)(reader->GetPixelRepresentation() * reader->GetRescaleSlope()) + (double) reader->GetRescaleOffset();


> // HuofAir represents -1000 HU
> HUofBone = HUofAir + 2250.0;  // 1250 HU for Bone
> HUofMuscle = HUofAir + 1050.0; // 50 HU for Muscle
> HUofFat = HUofAir + 880; // -120 HU for Fat
>


How are you reading your images? They already should be in Hounsfield

( no no calculation is needed) if you use the vtk image readers and
use signed short images.

John



      
_______________________________________________

Powered by www.kitware.com



Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html



Please keep messages on-topic and check the VTK FAQ at: http://www.vtk.org/Wiki/VTK_FAQ



Follow this link to subscribe/unsubscribe:

http://www.vtk.org/mailman/listinfo/vtkusers





-- 
Wesley D. Turner, Ph.D.
Kitware, Inc.
Technical Leader
28 Corporate Drive
Clifton Park, NY 12065-8662
Phone: 518-881-4920





      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20100826/32c670dc/attachment.htm>