1

I have a question pertaining to the conversion of patient coordinates into voxel coordinates of a 3D MRI point cloud model obtained from DICOM. I have a 3D point cloud file obtained from 2D MRI images, and the points are in the patient coordinate system. I am trying to convert those points in the cloud into voxel coordinates using the inverse of an affine matrix that I had initially created. But the affine matrix I have is singular and hence I cannot directly apply the inverse function on it. Is there a way to either construct an inverse affine matrix without applying the inverse function on the affine matrix? Or is there any other way to implement this conversion? I have also tried to get the pseudo-inverse matrix from the affine matrix, but I am not sure if that would help in any way.

I am working in Python 3.7. The shape of the affine matrix is (160, 4, 4), that of the pseudo-inverse matrix is also (160, 4, 4), and that of the point cloud array is (2086604, 4).

To convert to voxel coordinates, the code and resulting shape of the array are as follows:

voxels = np.dot(p_inverse_whole, stack_patient.T)  
(160, 4, 2086604)

Ideally I would like to have the resulting array in the following shape: (160, 4, 4) instead of (160, 4, 2086604)

Could anyone of you please help me solving this problem?

Amit Joshi
  • 15,448
  • 21
  • 77
  • 141
Ashritha Eadara
  • 11
  • 2
  • @Ashrita Eadara I'm facing a related problem: https://stackoverflow.com/questions/71471317/how-to-use-the-dicom-affine-transformation-matrix-to-convert-patient-coordinates Did you manage to find a solution? – lbrandao Mar 14 '22 at 16:46

0 Answers0