MRI Derived CT Substitute

Adam Johansson¹, Joakim Jonsson¹, Mikael Karlsson¹, Tufve Nyholm¹

A method using a Gaussian mixture model (GMM) for generating a substitute for a CT image from a set of MR images is presented. The heads of five patients were imaged with CT, a T2 weighted sequence and two dual echo UTE sequences employing both different echo times and flip angles. The method is evaluated using leave-one-out cross-validation. The mean absolute error of the CT number in the substitute CT images compared to the real CT images is found to be 141 Hounsfield units.