Application of nonlinear system identification to magnetic resonance imaging and computed tomography

The authors have examined the nonlinear behavior of MRI and CT using the Wiener-Volterra model of nonlinear systems, which allows evaluation of the type and quantitative significance of nonlinearities. The calculated nonlinear kernels revealed at least second order quantitatively significant nonlinearities in both CT and MRI related to spatial factors, such as geometric distortions and edge effects, and grey map transformations between modalities. Kernel calculations were complicated in CT image transformation mappings by partial volume effects, and in MRI mappings by magnetic susceptibility related image distortion.