Prefrontal gamma oscillations reflect ongoing pain intensity in chronic back pain patients

Figures

• 

  Figure 1. Single‐subject spontaneous pain ratings. Individual pain ratings are shown for the analyzed time window. Pain intensity was continuously rated on a visual analogue scale anchored at no pain and worst imaginable pain. Ratings were smoothed using a moving average with a window length of 1 s and a step size of 0.1 s. Ratings were subsequently z‐transformed to account for varying strength of pain rating fluctuations across participants.

• 

  Figure 2. Relationships between ongoing pain intensity and neuronal oscillations on electrode level. (a) Electrode level t maps of the relationship between ratings during the spontaneous pain and visual control condition and brain activity as assessed by linear regressions for theta (4–7 Hz), alpha (8–13 Hz), beta (14–29 Hz), and gamma (60–90 Hz) frequencies. Scaling reflects t values resulting from nonparametric cluster‐based permutation tests. Positive and negative relationships are reflected by warm and cold colors, respectively. Electrodes within significant clusters are marked. n.s., not significant; **p < .01 (two‐sided). (b) The right panel descriptively displays the frequency spectrum of the relationship between pain intensity and brain activity in the spontaneous pain and visual control condition for electrode Fz, which is highlighted in the topography in the left panel. Again, t values are shown. Frequency bands used in all analyses are marked. The strongest (positive) relationship was observed at 70 Hz. Relationships between 45 and 55 Hz, which are contaminated by line noise artifacts, are masked. [Color figure can be viewed at http://wileyonlinelibrary.com]

• 

  Figure 3. Trend analysis of the relationship between ongoing pain intensity and neuronal oscillations in the gamma frequency band. Box plots of individual, z‐transformed gamma activity at electrode Fz in the spontaneous pain condition are shown, sorted into five bins based on pain ratings. Gamma activity increased with increasing pain. A repeated measures ANOVA revealed a significant main effect of the bin number (F (3, 80) = 7.23, p < .001). A subsequent trend analysis showed a significant linear trend (F (1, 30) = 13.62, p = .001), while quadratic, cubic, and quartic trends were not significant (quadratic: F (1, 30) = 1.13, p = .30; cubic: F (1, 30) = .11, p = .74; quartic: F (1, 30) = .48, p = .49). rmANOVA, repeated measures analysis of variance.

• 

  Figure 4. Relationships between ongoing pain intensity and neuronal oscillations on source level. Source‐level t maps of the relationship between ratings during the spontaneous pain and visual control condition and brain activity as assessed by linear regressions for theta (4–7 Hz), alpha (8–13 Hz), beta (14–29 Hz), and gamma (60–90 Hz) frequencies. As in the previous figure, scaling reflects t values resulting from nonparametric cluster‐based permutation tests and positive and negative relationships are reflected by warm and cold colors, respectively. In plots showing significant relationships, areas outside of significant clusters are masked. in plots without significant effects, opacity is reduced. n.s., not significant; *p < .05 (two‐sided), **p < .01 (two‐sided). [Color figure can be viewed at http://wileyonlinelibrary.com]

• 

  Figure 5. Control analyses of muscle activity. (a) Relationships between pain ratings and activity in the gamma frequency band (60–90 Hz) in the spontaneous pain condition are shown for the exemplary fronto‐central EEG electrode Fz (see left panel in Figure 2b) and two EMG electrodes placed on the right masseter and neck muscles. As in previous figures, t values resulting from nonparametric cluster‐based permutation tests based on linear regressions are shown. EEG, electroencephalography, EMG, electromyography, n.s., not significant; ***p < .001. (b) Source‐level t map of the relationship between ratings during the spontaneous pain condition and data reconstructed from independent components, which were classified as artifact‐contaminated during preprocessing. Relationships based on linear regressions are shown for the gamma frequency band (60–90 Hz). As in previous figures, scaling reflects t values resulting from nonparametric cluster‐based permutation tests, positive and negative relationships are reflected by warm and cold colors, respectively, and opacity is reduced as no significant effects were found. n.s., not significant. [Color figure can be viewed at http://wileyonlinelibrary.com]