Investigation of DTI Bootstrap Reproducibility in the Human Brain

Robert Stefan Vorburger¹, Carolin Reischauer¹, Katerina Dikaiou¹, Philipp Staempfli¹, Peter Boesiger¹

Bootstrap methods were recently introduced in diffusion tensor imaging (DTI) to quantify the uncertainty in derived parameters. In this study, four different bootstrap algorithms were investigated with respect to their reproducibility in in-vivo measurements. Therefore, the bootstrap techniques were applied to selected regions of interest in the human brain. The reproducibility was demonstrated in a clinical measurement setup in-vivo. A difference concerning the reproducibility between the model based and non-model based techniques cannot be distinguished. Therefore, the model based bootstrap methods provide powerful techniques to quantify the uncertainty of DTI parameters and offer supplementary information without requiring additional measurements.