[Gate-users] Material image dose scoring

Fri Jun 22 10:02:23 CEST 2018

Dear Maikol/GATE-Users,

Please, I followed your prescribed method and worked successfully. However, the importation of the behaviour of the mouse mesh volume(bones) in the mother volume is unexpected. I expected the size of the mouse mesh volume to be smaller and located inside the mother volume. Attached is the picture of how the entire geometry looks like. Please, can anyone help me rectify this problem? Thanks for your help so far.

Best regards,

Emmanuel

________________________________
From: Emmanuel Marfo
Sent: Monday, 18 June 2018 9:05:05 PM
To: Maikol Salas Ramirez
Subject: Re: [Gate-users] Material image dose scoring

Thanks very much for your response. I will work on it and let you know the outcome.

best regards,

Emmanuel

________________________________
From: Maikol Salas Ramirez <mmsalas at gmail.com>
Sent: Monday, 18 June 2018 8:58:51 PM
To: Emmanuel Marfo
Subject: Re: [Gate-users] Material image dose scoring

Hi,

I just realize that just with the first step you will have the water+bone+fat mask because all the voxel most have to be 1 and air will be zero or NaN (1 after fix).

Now, with your question, the idea is to do not leave gaps. You start in step 1 with a full mask, then step by step you remove bone and fat, water can be considered a common material.
If you just get the water material you will have to apply a threshold value for water and you could have gaps between volume.

In step 3, you will have to select a threshold value for bone, with this value you will decide what is bone and what is not bone in the bone image, you will lost some info but it is the only way if you want to segment the would body.

In step 4, again you will have to select a threshold value for fat, with the same implication.

In step 5, the rest in the mask of step 2 will be water.

You also can try to start with water instead of bone, but I found that bone is better for the high contrast and fat second, because fat has a more specific localization in the body.

Best regards
Maikol

2018-06-18 10:35 GMT+02:00 Emmanuel Marfo <emmanuel.marfo at postgrad.otago.ac.nz<mailto:emmanuel.marfo at postgrad.otago.ac.nz>>:

Thanks for the reply. This process below is for me to make sure I understood the method clearly and ask a question another question. Perhaps I will also be able to know the answer through some measurements

First step

  1.  Obtain mass density probabilities for the materials concerned (such as bone, fat and water). The summation of their probabilities should be 1. Fix all NaN.

Second step

2. Create a mask from the summed materials (wate+fat+bone)

3. Create a mask and a mesh volume from the bone. Subtract the bone mask from (water+fat+bone) mask leaving (water+fat )mask

4 Create a mask and a mesh volume from the fat. Subtract the fat mask from (water+fat) mask leaving pure water mask

5. Create a mesh volume of water from a water mask.

6. Now, we have three masks and three meshes of water, bone and fat materials. The water mesh volume serves as the mother volume while the bone and fat mesh serves as the daughter volumes.

The question with the pure water mask, why do we have to add water,fat and bone mask,  and  then subtracting fat and bone mask to get pure water mask? Can`t we just get the water mask and mesh volume from the water mass fraction image.

________________________________
From: Maikol Salas Ramirez <mmsalas at gmail.com<mailto:mmsalas at gmail.com>>
Sent: Monday, 18 June 2018 7:58:44 PM

To: Emmanuel Marfo
Subject: Re: [Gate-users] Material image dose scoring

Hi Emmanuel,

Yes, you are right.
There is a possibility to have some voxels with pure bone, therefore it is better to include bone in the principal mask to have a complete volume.
Sorry, I did not see that.

Best regards.
Maikol

2018-06-18 9:36 GMT+02:00 Emmanuel Marfo <emmanuel.marfo at postgrad.otago.ac.nz<mailto:emmanuel.marfo at postgrad.otago.ac.nz>>:

Hello Maikol,

Please, I have a question about subtracting bone and fat mask from water + fat maks to get pure water mask. Why are we not summing water, fat and bone mask and then subtract bone and fat mask from them but instead water + fat mask only?

Best regards,

Emmanuel

________________________________
From: Maikol Salas Ramirez <mmsalas at gmail.com<mailto:mmsalas at gmail.com>>
Sent: Friday, 15 June 2018 12:03:46 AM
To: Emmanuel Marfo
Subject: Re: [Gate-users] Material image dose scoring

Hi Emmanuel,

I am not an expect, but I will recommend some possible steps:

1- You sum the fat and water images and than transform the image in a mask, I mean all the voxel with a value higher than zero, yo set them to 1.

2- Segment the bone image, for example you can use a threshold, which all the voxels with a value higher than 0.1 (this value you have to decided) in the bone image will assigned a value of 1, then you will have a bone mask, then you subtract it from the first mask (step 1). At same time you use the bone mask to create a mesh volume, you can do this procedure in 3DSlicer or in ImageJ (using 3d viewer).

3- in this point you have a three masks (water + fat, bone, water + fat - bone ) and one mesh volume (bone)

4- You take the fat image, and you make a mask of this image (again you have to chose a threshold value). Also you can make a mesh volume of this mask.

5- Now you substract from the  water + fat - bone mask the fat image, then you will have a mask of pure water. Then you can create a new mesh for water.

You can do all of this steps with ImageJ, and it is very important that each time that you create a new mask, you have to apply some filter to smooth the mask and avoid holes (please see the effect over the image of the imageJ funtion in Process->Binary and Process->filter), also go through all the images to check the mask. Only after this fine tune you have to do mask subtractions.

At end you will have three mesh files and thre masks. The water-mesh in the mother volume and the fat an bone are child volumes.

Let me now if it works. I hope, It will help you.

Best regards
Maikol.

2018-06-14 11:48 GMT+02:00 Emmanuel Marfo <emmanuel.marfo at postgrad.otago.ac.nz<mailto:emmanuel.marfo at postgrad.otago.ac.nz>>:

Dear Sir,

Please, I followed your procedure and was able to convert my images to mass density fraction for lipid, water and HA. I was also able to correct the NaN too.

Now I am wondering what the next step will be. Please, can I adopt your method for the MRI image(lipid/water) and see if it can work for me. The procedure you talk about by creating a mask of each material or creating mesh volumes for each material is not quite clear to me. Can you help me with that? I have attached an interface of a plugin in ImageJ called morphological segmentation showing one of my images. I was hoping this could help me create the mask of each material. I am not sure if is the right method. I am not an expert in Imaging. I seek your guidance.

>>I think, it is not possible to use the material decomposed images directly in GATE without a table,

  1.  one option is to create a mask of each material and
  2.  A second is to create mesh volumes of each material and used them like a child volume of the contour of the normal ct image.

thanks.

Best regards,

Emmanuel

________________________________
From: Maikol Salas Ramirez <mmsalas at gmail.com<mailto:mmsalas at gmail.com>>
Sent: Wednesday, 13 June 2018 12:47:57 PM
To: Emmanuel Marfo
Subject: Re: [Gate-users] Material image dose scoring

Hi Emmanuel,

I think, it is not possible to use the material decomposed images directly in GATE without a table, one option is to create mask of each material and a second is to create mesh volumes of each material and used them like  a child volumes of the contour of the normal ct image.

One option to start is create fraction images:

When you use material decomposed images, you have mass conservation (it depend of your spectral method, but normally is like that, you have to check it), it means Mass_Fraction(HA) + Mass(Lipid) _Fraction + Mass(Water)_Fraction = 1.

The problem here is how to transform your images in fraction images, I had a previous experience but with Water/Fat images but from MRI. For my MRI experiment I calculated the fat fraction and the water fraction (1-fat fraction).

In your case you can do the same:
1- HA_Fraction = HA_Voxel_Value / ( HA_Voxel_Value +  Lipid_Voxel_Value +  Water_Voxel_Value).
2- Lipid_Fraction = Lipid_Voxel_Value / ( HA_Voxel_Value +  Lipid_Voxel_Value +  Water_Voxel_Value).
3- Water_Fraction = Water_Voxel_Value / ( HA_Voxel_Value +  Lipid_Voxel_Value +  Water_Voxel_Value).

I did this calculation with your images using ImageJ and it works pretty easy, you just need to sum all the three images and divide each one by the sum, if you chose one voxel you will see that the sum is 1. The attached image shows the three images.

The only artifact here is in voxel with air or with a zero value you will have NaN value, you have to fix them (given the zero value), ImageJ has an option to do that.

--------------
This is what I can see, the main point is that a normal voxel in a CT images has a composition of material and what you have in each voxel is a effective atomic number and effective density, with spectral CT you separate the material and you have to take care to use the individual images because you lost anatomical information information.

Sorry I think, I could not help you to much.

Best regards
Maikol

2018-06-12 10:50 GMT+02:00 Emmanuel Marfo <emmanuel.marfo at postgrad.otago.ac.nz<mailto:emmanuel.marfo at postgrad.otago.ac.nz>>:

Hello

Thanks for the quick response. Please, the images are density images obtained by dividing linear attenuation(energy information) values by mass attenuation(material basis).

Attached is a dropbox link of a folder containing the material decomposed images for your assessment. thanks

https://www.dropbox.com/sh/zxfio5azk4q6osb/AADw0a1BMYQiPnqFjMgwdCnYa?dl=0

[https://cfl.dropboxstatic.com/static/images/icons128/folder_dropbox.png]<https://www.dropbox.com/sh/zxfio5azk4q6osb/AADw0a1BMYQiPnqFjMgwdCnYa?dl=0>

material images from spectral CT<https://www.dropbox.com/sh/zxfio5azk4q6osb/AADw0a1BMYQiPnqFjMgwdCnYa?dl=0>
www.dropbox.com<http://www.dropbox.com>
Shared with Dropbox

Best regards,

Emmanuel

________________________________
From: Maikol Salas Ramirez <mmsalas at gmail.com<mailto:mmsalas at gmail.com>>
Sent: Tuesday, 12 June 2018 5:45:25 PM
To: Emmanuel Marfo
Cc: gate-users at lists.opengatecollaboration.org<mailto:gate-users at lists.opengatecollaboration.org>
Subject: Re: [Gate-users] Material image dose scoring

Hi Emmanuel,

When you said material decomposed image, are you talking about Rho/Z image (effective density and atomic number - image)?
What info do you have in your image?

Best regards
Maikol

El El mar, 12 jun 2018 a las 0:34, Emmanuel Marfo <emmanuel.marfo at postgrad.otago.ac.nz<mailto:emmanuel.marfo at postgrad.otago.ac.nz>> escribió:

Hello Gate Users,

Please, I need help on this or if anyone has the knowledge of how it can be done I would be grateful. I am using a spectral CT system to obtained material decomposed image based on mass attenuation and energy information. I want to do dose deposition on the material decomposed images without the use of Schneider material converter employed in GATE.  My problem is I want to know if GATE architecture supports it and second I can not import the material decomposed image into GATE because is always requesting for a conversion text file. Can I bypass it? Can someone help me, please? Thanks

Best regards,

Emmanuel Marfo

Student

University of Otago

_______________________________________________
Gate-users mailing list
Gate-users at lists.opengatecollaboration.org<mailto:Gate-users at lists.opengatecollaboration.org>
http://lists.opengatecollaboration.org/mailman/listinfo/gate-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opengatecollaboration.org/pipermail/gate-users/attachments/20180622/55741da1/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Image2.png
Type: image/png
Size: 20892 bytes
Desc: Image2.png
URL: <http://lists.opengatecollaboration.org/pipermail/gate-users/attachments/20180622/55741da1/attachment-0001.png>