Hi,

There are lot of scripts to do this. See for example dcm2nii or r2agui or SPM dicm_convert...

Try MeVisLab (http://www.mevislab.de/) to start with.

Hi,

I would like to rephrase your words in to 3D volume instead of 3D image, as using the word 3D image may not be relevant here. Because ultimately we are going to represent the 3D volume (created from the set of 2D DICOM CT images) again through a 2D image using projections

You can follow other answers from the tools perspective and from the core logic perspective, see the explanation below.

In volumetric scan of patient, whatever the CT images we see they are axial cross sections represented in image coordinate system (ICS). And in practical the patient under scan is represented in DICOM Patient Coordinate System (PCS). In principle, we need to convert the representation of information from ICS to PCS before proceeding further. The size of pixel has its width in x-axis and its height in y-axis. If we consider the depth of the volume in z direction, we come across slice location. With the combination of these three, we get the volumetric pixel in 3D i.e voxel.

Before conversion from 2D to 3D, you need following DICOM tag values

1.       2D input point (x,y)

2.       Image position patient (0020, 0032),

3.       Pixel spacing (0028, 0030),

4.       Row vector and column vectors (0020,0037). These are also called as Image orientation patient.

Conversion from 2D to 3D

Voxel (x, y, z) = Image Plane Position + Row change in X + column change in Y

Where Row change in X = Row vector * Pixel size in X direction * 2D Point location in X direction

Column change in Y = Column vector * Pixel size in Y direction * 2D Point location in Y direction

For example, if Image position patient is (100,100,50), row vector=(1,0,0), column vector=(0,1,0), 2D point=(5,6) and pixel size=(0.5,0.5), then,

Voxel (x,y,z) = (100,100, 50) + (1,0,0) * 0.5 *5 + (0,1,0)*0.5* 6 =  (100,100,50) + (2.5,0,0)+ (0,3,0) = (102.5, 103, 50)

Conversion from 3D to 2D

Difference = Voxel – Image plane Pos;

Difference_in_X = Difference.Innerproduct(Row vector)

Difference_in_Y = Difference.Innerproduct(Col vector);

2D Point = (Difference_in_X/pixel size in X, Differnce_in_Y/pixel size in Y)

Finally you will get the same 2D point = (5,6)

This is the basic logic involved. The above set of equations can be realized through matrix multiplication also. Refer to page 410, DICOM chapter PS 3.3, 2011 version or else search for the tag 0020,0037) in the file http://dicom.nema.org/medical/dicom/current/output/pdf/part03.pdf

I hope this answers your question.

- Manjunath

Hi, have a look at the ITK and VTK libraries. ITK provides functions to read DICOM files, and VTK can be use for visualization. You'll find well explicated examples on how to use these libraries to reconstruct a 3D volume from a DICOM series.

Hope it helps.

Greg

http://itk.org/

http://www.vtk.org/

Hi, have a look at the ITK and VTK libraries. ITK provides functions to read DICOM files, and VTK can be use for visualization. You'll find well explicated examples on how to use these libraries to reconstruct a 3D volume from a DICOM series.

Hope it helps.

Greg

http://itk.org/

http://www.vtk.org/

Hi, have a look at the ITK and VTK libraries. ITK provides functions to read DICOM files, and VTK can be use for visualization. You'll find well explicated examples on how to use these libraries to reconstruct a 3D volume from a DICOM series.

Hope it helps.

Greg

http://itk.org/

http://www.vtk.org/

Hi, have a look at the ITK and VTK libraries. ITK provides functions to read DICOM files, and VTK can be use for visualization. You'll find well explicated examples on how to use these libraries to reconstruct a 3D volume from a DICOM series.

Hope it helps.

Greg

http://itk.org/

http://www.vtk.org/

Take a look at our Checkpoint software. It can load DICOM scans and perform 3D rendering as well. It also provides tools to collect 3D landmark points to perform 3D shape analysis.

https://www.stratovan.com/products/checkpoint?utm_source=rg&utm_medium=ans&utm_campaign=3dvis9

I am doing my MCA project on DICOM image viewer. How can I get the sagittal and coronal image views? I need the algorithm in C++? Kindly help.