Connectivity differences between Gulf War Illness (GWI) phenotypes during a test of attention

Figures

• 

  Fig 1. Selection of optimal Cohen’s d. (A) Cross-sections through the blue cone show the number of significant edges with FDR<0.01 at different levels of Cohen’s d. The Venn diagrams show the numbers of edges in SC (white), START (green), STOPP (red), SC&START (aqua), SC&STOPP (yellow), START&STOPP (orange) and shared by all 3 groups (black) at each level of d. (B) Jaccard indices show the similarity of each pair of groups. Similarity was lowest at Cohen’s d = 1.6. (C) The disorder, or absence of overlap between groups, was confirmed by the peak in Shannon’s entropy at the same level.

• 

  Fig 2. 0-back connectivity in SC. All edges with d>1.6 and FDR<0.01 were plotted as a ball and spring diagram. The dotted line separated task and default modules based on the Louvain method [79] weighted by Fisher’s z-transformed Pearson correlation coefficients. Edges that were significant exclusively in SC were depicted by thick blue lines. Dashed black lines indicated edges shared by all 3 groups. Thin black lines were shared by pairs of groups. Yellow colored nodes had betweenness centrality >0.15 or were of special functional significance in SC. Edge betweenness centrality >0.15 was indicated. Grey shaded areas indicated medial parietal, lateral parietal, DAN, RDLPFC, left ventrolateral prefrontal cortex, basal ganglia and anterior insulae nodes.

• 

  Fig 3. Comparison of bridging nodes and edges between groups. Modularity (Q) was calculated for default and task communities in (A) SC, (B) START, and (C) STOPP. They had 3 patterns of connecting links between DAN nodes. The arrangements of DAN and posterior salience (SP) nodes and edges were different between groups.

• 

  Fig 4. 0-back connectivity in START. All edges with d>1.6 and FDR<0.01 for START were plotted as described for Fig 2. The shaded nodes had particular relevance for the START group. The RE1—PD4 edge had high betweenness centrality (0.16). The dotted line separated task and default modules.

• 

  Fig 5. 0-back connectivity in STOPP. All edges with d>1.6 and FDR<0.01 for STOPP were plotted as described for Fig 2. The shaded nodes had particular relevance for the STOPP group. High edge betweenness centrality was indicated for two edges from RE2. The dotted line separated task and default modules.