Hello, we are using the dcm2niix conversion tool and trying to reverse the effects of the gantry tilt correction for overlaying some processed information onto the original dicom. However for working on the data, we use the tilt corrected output. I have tried using the information of the slice thickness to calculate the new sizes but cannot figure it out. here is the relevant data:

dcm: 512x512x158; slice thickness: 0.9
gantry tilt: -18.5
tilt_corrected_nii: 512x617x158; slice thickness: 0.9483

I think you have two options: 

1.) Normalize the tilt-corrected image to the raw (distorted) image and apply the transform to the processed images. Choose a normalization tool that allows you to apply a full affine (12 parameter model) transform (9 parameter transforms are insufficient, you are correcting for a shear). FSL's FLIRT would be an excellent tool for this. 

2.) Write a script that removes the shear from the tilt-corrected image. The function is described on these pages, as well as the dcm2niix code:

https://github.com/rordenlab/dcm2niix/is...

https://raw.githubusercontent.com/rorden...

float GNTtanPx = tan(gantryTiltDeg / (180/M_PI))/hdrIn.pixdim[2]; //tangent(degrees->radian)