Examining connectivity using functional magnetic resonance imaging (fMRI) is an important method to understand brain functions. However, previous investigations mainly relied on static approaches to study functional connectivity (FC). With recent advances in simultaneous multi-slice imaging⁽¹⁾, which have improved the temporal resolution of fMRI, it is possible to capitalize on the information contained within the temporal features of BOLD FC using a sliding window approach⁽²⁾. As we plan to employ this technique for evaluation of brief, spontaneous events (EEG-spikes in epileptic patients), we aimed to determine the effect of different window sizes on the temporal dynamics. Here, we investigated to which extent window sizes of 10, 20, and 30 samples (6.6, 13.2, and 19.8s) can capture the dynamic network changes that occur during visual stimulation (checkerboards).5 healthy volunteers (mean age 28 ± 2.0 years) participated in this fMRI study comprising four imaging runs (each of 10 minutes) after providing informed consent. Images were acquired using a Multiband (MB) EPI sequence with isotropic resolution of 3 mm (40 slices) at a sub-second sampling rate (TR = 0.66s, MB factor of 4) using a 3T MRI system (Magnetom Prisma, Siemens, Germany) equipped with a 64-channel head coil. The fMRI data were recorded from a simple checkerboard visual experiment (block design) where the duration of the checkerboard stimulation was varied from 5, 10, 20 and 30 seconds across the four imaging sessions. The order of these runs was counterbalanced across subjects. Each imaging session (10 minutes with a total of 910 EPI volumes) comprised of three repetitions of the checkerboard stimulation. In all imaging sessions, subjects were instructed to focus on the fixation cross at the center of the screen and press the button response with their right thumb whenever a circular checkerboard was presented in order to remain vigilant during the multiple imaging sessions. Preprocessing relied on FSL, SPM8 and REST. The sliding window analysis was performed using the DynamicBC toolbox⁽³⁾ with varying window sizes (1 data point shifts). The toolbox is based on Pearson linear correlation to compute bivariate FC between every pair of voxels. Subject specific whole brain gray matter masks (SPM12, unified segmentation) at a probability threshold of 0.2 were used as the ROI for the connectivity analysis. We estimated the functional connectivity degree (FCD) that counts the total number of connections of a given voxel above a predefined threshold (Pearson linear correlation, p<0.001). The FCD values were averaged across the three repetitions of the task (20s before to 100s after the task), and the resulting values were averaged across subjects for each stimulation length, and plotted for regions defined by the Freesurfer Desikan-Killiany atlas.Figure 1 shows the temporal dynamics for the 29 regions. Highest FCD values were observed in the lateral occipital region (overlapping with V1), which was used as the ROI to plot the FCD at different window sizes and varying block lengths (Figure 2). With respect to window size, highest FCD values (Figure 2) were observed for the 19.8s window (30 samples). Figure 2 also shows a double peak related to the stimulation duration. Figure 3 shows the percent difference in mean FCD between task (30s after stimulus) and rest (20s before stimulus) over different window sizes and stimulation lengths. The largest percent difference (compared to baseline) in FCD was observed for the 19.8s window at a stimulation length of 10s.The increasing difference of the observed double peak is an indication of increasing stimulation block durations. Concerning the effect of the window size on the FCD, maximum percent difference of 54% was observed for 19.8s window, probably because larger window sizes include more samples and, consequently, can provide a more robust and more significant estimation of the correlation coefficients. However, larger windows will also introduce temporal smoothness that needs to be balanced with the advantage of a more robust FCD estimation. Based on the percent difference (task vs. rest) results we can conclude that larger windows of 13.2 and 19.8s provide a more robust estimation of FCD in contrast to the 6.6s window which showed a maximum percent difference of only 18%. But due to the higher temporal smoothness of the 19.8s window (high FCD values lasting around 40s after onset as compared to 30s for the 13.2s window at the same stimulation length of 30s), 13.2s window would be an optimum choice among the windows used in the current setup. These results are encouraging and but require confirmation in a larger group and patients with epileptic spikes.