UMCG/RuG BCN - NIC Journal club 25 Apr. ’08 A method for functional network connectivity among spatially independent resting-state components in schizophrenia Jafri et al., NeuroImage 39 (2008)
BCN NIC Introduction Functional connectivity methods Seed-voxel functional connectivity mapping Independent component analysis Functional network connectivity: time dependence among ICA components Applicable to cognitive or motor tasks, or resting state Group comparisons Schizophrenia
BCN NIC Materials & Methods Subjects 29 schizophrenics 25 matched healthy controls Scanning 3.0-T GE-EPI fMRI TR = 1.86 s 162 acquisitions [!] Preprocessing Motion correction Spatial smoothing, FWHM = 10 mm [?] Normalization to MNI, conversion to T&T
BCN NIC Materials & Methods Group-sICA All subjects pooled, verified per group InfoMax algorithm 30 components extracted Temporal concatenation and back-reconstruction 7 components systematically [?] selected on the basis of correlations with CSF and GM
BCN NIC Results
BCN NIC Materials & Methods Correlation and lag analysis Band-pass filtering at f = Hz (i.e., T = s) Interpolation to higher time resolution [?] For each of 21 pairs of components, determined maximal correlation coefficient ρ and corresponding time lag δ (−5 s < δ < 5 s) Significant correlations extracted using t-test at p < 0.05 Group comparisons using conservative t-test at q < 0.05 Resampling by subsampling groups Resampling by relabeling groups
BCN NIC Results
BCN NIC Results 50% 60% 50% 55% 65% 70% 65%
BCN NIC Remarks Network connectivity methods are feasible Significant outcomes in individuals Significant differences between groups The method can be expanded SEM GCA EEG Statistics remain doubtful Can t-tests be performed on correlation coefficients ρ? How did corrections for multiple comparisons take place?