Efecto del tamaño kernel en el suavizado de señal BOLD en paradigmas funcionales (RMf)

Smoothing is a filtering technique that is essential for brain signal analysis and consists in calculating and comparing the average activation of a voxel to that of its neighbours. Several authors have proposed alternatives or modifications to this process; nonetheless, articles that compare the effect of different sizes of smoothing remain scarce. Thus, the aim of this study was to investigate the effect of applying different smoothing sizes and to highlight the importance of choosing the correct smoothing size. Five smoothing criteria were applied to brain images obtained during an easy motor task performed by five adult participants. Significant differences were found between different smoothing sizes, mainly between the non-smoothing application and the smallest smoothing size versus the two largest smoothing sizes. The signals from the most activated brain areas did not disappear with increased smoothing, whereas signals from less active or smaller areas disappeared. Despite the study sample size, the results suggest that smoothing is relevant in functional magnetic resonance image processing and that the optimum smoothing size is 2.5 and 3