72 CHAPTER 4 analysed the influence of between-group factor GROUP, and repeated factors POSTURE (congruent, incongruent) and LATERALITY on median RT and normalized ER with two additional 3-factor mixed ANOVAs. To correct for multiple testing (4 ANOVAs) we set alpha at p = 0.0125 for these analyses. Task-related cerebral activity Image analyses were performed using FSL version 5.0.11. 129 On the first (subject-specific) level, we used a general linear model (GLM) to model blood-oxygen-level dependent (BOLD) activation per participant. 139, 140 Task design matrices and contrast images of parameter estimates of the BOLD signal were generated using FEAT. 131 Our statistical model at the first level included the factors LATERALITY (left, right), BIOMECHANICAL COMPLEXITY (easy, complex), and POSTURE (congruent, incongruent). We additionally included interaction terms for LATERALITY x BIOMECHANICAL COMPLEXITY and LATERALITY x POSTURE, to investigate the limb-specific effect of movement complexity and somatosensory changes on activation, respectively. All regressors of interest included correct responses only, and events were time-locked to stimulus onset, with a duration corresponding to the median RT over all trials per participant, following the same approach as in previous work in healthy and clinical populations. 30, 40, 58, 59, 141 The model additionally included 3 regressors of non-interest: between block hand re-positioning (7 seconds), incorrect trials and missed trials (with median RT durations). All regressors were convolved with a canonical hemodynamic response function (HRF) and its temporal derivative to model BOLD activation. 142 FMRIB’s Improved Linear Modelling (FILM) prewhitening was performed before voxel-wise fitting of the GLM. 131 Group analyses were done using FSL’s randomise tool143 to perform non-parametric threshold-free cluster enhancement (TFCE) based permutation testing (5000 permutations), correcting for multiple comparisons with a Family Wise Error (FWE) of p < 0.05. For each contrast of interest, the GLM consisted of factor GROUP (NA or healthy), with the individual contrast of parameter estimate images from the first level analyses as input. Contrasts of interest were LATERALITY (left > right, right > left), BIOMECHANICAL COMPLEXITY (complex > easy) and POSTURE (incongruent > congruent), as well as their interaction terms (rightcomplex>easy > leftcomplex>easy , rightincongruent>congruent > leftincongruent>congruent). We additionally looked for shared activity and general task effects by running a single group simple design for the contrasts relating to factors LATERALITY (left > right, right > left), BIOMECHANICAL COMPLEXITY (complex > easy, collapsed over LATERALITY and for left and right hands separately) and POSTURE (incongruent > congruent). Post hoc functional connectivity analysis As NA patients showed altered brain activity compared to healthy participants during mental rotation of right (affected) versus left (non-affected) hands in two brain regions (see Results), we performed post-hoc seed-based functional connectivity analyses to explore the underlying networks. To this end, we extracted the mean BOLD signal within those regions (MNI coordinates peak voxel, right extrastriate cortex: [50 -66 12], 114 voxels; bilateral parieto-occipital sulcus: [10-58 16], 429 voxels; see Table 3) and ran two separate GLMs with the timeseries of each of the regions as a single regressor. For each of the regions, we compared the resulting contrast of parameter estimate images between
RkJQdWJsaXNoZXIy MjY0ODMw