[MINC-users] DICOM to MINC

Michal Povazan michal.povazan at meduniwien.ac.at
Tue Sep 24 06:24:14 EDT 2013


Hi all,

I would like to ask a question concerning the coordinate systems and 
transformation from DICOM to MINC

We are trying to transform the DICOM patient coordinate system into MINC 
world coordinates (our actual task is the creation of metabolic maps in 
MINC format from 3D CSI Siemens DICOM data).
How is this transformation carried out in MINC converter (dcm2mnc), or 
in other words how to transform the DICOM`s positions, sliceNormalVector 
and InPlaneRotation parameters to direction cosines and start values in 
MINC?
We were able to compute correct direction cosines, however, the start 
values are wrong for z direction. we have used the following code as an 
inspiration:

from dicom_read.c (dcm2mnc)

/for (ivolume=0; ivolume < VOL_NDIMS; ivolume++) {//
////
//         if (found_coordinate && //
//             found_dircos[VSLICE] && //
//             found_dircos[VROW] &&//
//             found_dircos[VCOLUMN]) {//
//             starts[ivolume] = //
//                 coordinate[XCOORD] * dircos[ivolume][XCOORD] +//
//                 coordinate[YCOORD] * dircos[ivolume][YCOORD] +//
//                 coordinate[ZCOORD] * dircos[ivolume][ZCOORD];//
//         }//
//         else {//
//             starts[ivolume] = 0.0;//
//         }//
//     }//
//
//
/and for the coordinates we have used (from siemens_to_dicom.c)

      for (i = 0; i < WORLD_NDIMS; i++) {
          coord[i] -=
              pixel_spacing[0] * ((double) ncolumns - 1.0) / 2.0 * row[i] +
              pixel_spacing[1] * ((double) nrows - 1.0) / 2.0 * column[i];
      }

The problem is, if I understand it correctly, this piece of code compute 
the coordinates just for slice and since we have 3D datasets, this can`t 
work properly.

What is the correct formula for computation of start values from 
position and direction cosines?

I hope my description is understandable.

thank you for any help

Michal Povazan


More information about the MINC-users mailing list