30. Mapping of movement artifacts in fMRI

Abstract

Listen

Translate article

Introduction Movement artifacts can cause serious changes in fMRI time series and therefore induce false positive or false negative results. In this work we calculated maps of movement artifacts using various approaches and compared maps across three different datasets to assess the variability between different populations and acquisition parameters. Methods Three different data sets (each under different experimental task, TR and number of scans, each consisted of 50 subjects), were included into the evaluation. The data was processed using SPM8. 6 parameters obtained during realignment were extended by adding their first differences and their squares to get up to 24 parameters included in the design matrix. Subsequently these movement parameters were tested for their statistical significance on explanation of variability in the data. Probability maps were created by calculating relative frequency of occurrence of “movement-based activation” within each dataset. Power maps were calculated from the ratio of movement-based signal energy to the energy of the total measured signal. Results In this work, we demonstrated that using all of 24 parameters will explain more variability in the data related to movement artifacts. We found out that probability maps show the similar location of artifacts across datasets nevertheless maps differ in frequency of occurrence. Power maps indicate similar places as probability maps but the amount of the movement-based variability is different across datasets. Conclusion We can conclude, that movement artifacts influence the signal at similar areas within subjects and datasets. Movement artifacts are mainly localized at the edges of the brain and brain ventricles. However, consistency of artifact occurrence and power of movement-based signal is different between studies.