TIGER - A New Model for Spatio-temporal Realignment of FMRI Data

We describe a new model which is able to model accurately the characteristics of subject motion, a dominant artefact in Functional Magnetic Resonance Images. Using the model, which is based on specific knowledge regarding the nature of the image acquisition, it is possible to correct for this motion which would otherwise render activation detection on the images invalid. We also present an initial implementation based on the model and are able to demonstrate that the corrections available under this new scheme are significantly more accurate than existing approaches to the problem of subject motion, enabling a far more accurate analysis of the patterns of brain activation which these images seek to capture.