# \[SOLVED\]Voxel to phisycal position in GPU
**URL:** https://discourse.itk.org/t/solved-voxel-to-phisycal-position-in-gpu/1011
**Category:** Uncategorized
**Created:** [June 16, 2018, 7:01pm UTC](https://discourse.itk.org/t/solved-voxel-to-phisycal-position-in-gpu/1011 "2018-06-16T19:01:43Z")
**Posts on this page:** 2
**Page:** 1
### Author:  [@rolof](https://discourse.itk.org/u/rolof)
#### Post date: [June 16, 2018, 7:01pm UTC](https://discourse.itk.org/t/solved-voxel-to-phisycal-position-in-gpu/1011/1 "2018-06-16T19:01:44Z")
Hi everyone,
I know that in CPU I have the containers to 3d images and I have available a container to work with GPU. However, I saw in all examples that GPU container examples only access to pixel value with a casting to float or the kind of image inside the kernel.
Now I need for different reasons work in GPU with ITK images containers and physical world values of a voxel. I CPU I know that I can use “TransformIndexToPhysicalPoint” function to do it. However, I need to do inside the GPU kernel the works that the function “TransformIndexToPhysicalPoint” does. I do not know how to do it, Can anyone help me?
On the other hand, If I have another small image that is a Region of interest of the first image and I want to obtain the physical point inside the GPU kernel, it is possible?
Thanks in advance.
---
### Author:  [@rolof](https://discourse.itk.org/u/rolof)
#### Post date: [June 29, 2018, 11:33am UTC](https://discourse.itk.org/t/solved-voxel-to-phisycal-position-in-gpu/1011/2 "2018-06-29T11:33:52Z")
The answer of this question can be found in another question that I did for CPU, here [TransformIndexToPhysicalPoint manually](https://discourse.itk.org/t/transformindextophysicalpoint-manually/1031)
So we only need to pass to GPU kernel the vectors or matrix to do the computation of the physical values inside the kernel.