Reliability of using a fixed matrix in coregistration of combined PET-MRI in a split magnet design

We consider the effects of using a fixed linear transformation to match positron emission tomography (PET) and magnetic resonance imaging (MRI) data acquired simultaneously using a split-magnet system. Estimates of the frequency offset in MRI scans were used to calculate geometric variability in MRI reconstruction as a consequence of mis-setting this parameter in addition to repeated estimation of the transformation matrix by manual measurements. None of the measured variability approached the resolution of the PET images so we concluded that a fixed matrix can be reliably used in such a system.