Is it possible to manipulate the orientation parameters?

[xfgavin]

For example, I would like to have my converted image to be in LPS orientation not in LAS.
Thanks.

[neurolabusc]

In theory you can edit nii_dicom_batch.cpp and change the line opts->isFlipY = true; to read opts->isFlipY = false; and recompile. This will change the order of the rows as written to disk. However, the purpose of this line is to emulate other DICOM converters, and not to ensure LPS vs LAS convention. For all 2D images my software will preserve the slice orientation as saved to disk in the DICOM file, therefore while rows are the P-A dimension for axial acquisitions they are the I-S dimension for coronal and sagittal scans. This is required for slice timing tools. If my software detects that an image is a 3D acquisition, it will reorient the volume to the closest orthogonal rotation to an NIfTI identity matrix [1,0,0; 0,1,0; 0,0,1] (though note many tools store their templates with the left-right reversed, e.g. [-1,0,0; 0,1,0; 0,0,1]).

In all cases, my software strives for lossless copying of raw image data (no interpolation) and the spatial transforms are stored in the header s-form and q-form. If you want to reorient your images, you can use fslreorient2std - though my caveats regarding slice timing should be considered. A much better solution is to let the viewing software take care of the rotations for you. This can optionally include lossy reorientation that gets the image aligned nicely without degrading the quality of the stored data.